Chapter 8: Data Analysis and SPSS Statistics
Using SPSS Statistics specifically, Chapter 8 focuses on data analysis and hypothesis testing in the context of teaching languages. The chapter discusses several data types (nominal, ordinal, interval, and ratio), how to interpret descriptive statistics, and various hypothesis tests, such as the independent samples t-test, paired samples t-test, one-sample t-test, correlation, and one-way ANOVA.
Overall, this chapter provides language teachers who want to undertake data analysis and hypothesis testing to enhance their teaching methods with a thorough manual. Step-by-step instructions on how to carry out each test using SPSS Statistics and how to interpret the findings are provided in this chapter. Examples of data tables are also included to demonstrate how the assessments may be used in language education research.
GUIDING QUESTIONS FOR DISCUSSIONS
- How crucial is data analysis in studies on language instruction?
- What are the various sorts of data that are often utilized in studies on language teaching?
- Which frequent statistical tests are used in studies on language education, and how?
- What are some possible drawbacks or difficulties of using statistical analysis in studies on language teaching?
INTRODUCTION
Quantitative research in language instruction requires important components like data analysis and outcomes presenting. Quantitative information is gathered for this kind of study by exams, surveys, observations, and experiments. Utilizing statistical techniques, this data will be analyzed to find trends, correlations, and patterns. The results must be communicated in a clear and simple way as part of the presentation.
A popular statistical technique for data analysis in studies on language instruction is descriptive statistics. The data gathered are summarized and described using descriptive statistics. You may compute measures like mean, median, mode, standard deviation, and range to provide an overview of the data. For instance, descriptive statistics were employed to describe the mean, median, and mode of the participants’ replies in a research that looked at ESL learners’ views on utilizing technology in language learning.
Inferential statistics are another statistical technique utilized in data analysis. On the basis of the gathered data, inferential statistics are used to test hypotheses and create predictions. Correlation analysis, ANOVA, and t-tests are examples of common inferential statistics. In a research that looked at the connection between Japanese English learners’ speaking abilities and their language anxiety, inferential statistics were employed to calculate the correlation between the two variables. It is crucial to express quantitative research outcomes in language training using clear and succinct language. The data may also be represented graphically using tables, graphs, and charts.
Reporting findings and interpreting the data are essential parts of quantitative research in language instruction. To ensure that the results are properly conveyed, the findings should be presented in a clear and succinct manner. Tables, graphs, and charts may be used to graphically depict the data, and descriptive and inferential statistics are often used in data analysis.
CASE STUDY
Case Study: Examining a Vocabulary Learning Strategy’s Efficacy
English instructor Mr. Brown wants to assess the success of a vocabulary-learning technique he has been using with his pupils. He has data from his lessons, and he wants to examine it to see whether the method has helped his kids learn more vocabulary. He intends to utilize SPSS Statistics to carry out numerous hypothesis tests for this.
Tasks:
- What kind(s) of information should Mr. Brown gather in order to assess the potency of the vocabulary-learning strategy?
How can Mr. Brown manage and input his data into SPSS Statistics?
- How can Mr. Brown understand the descriptive statistics he uses to examine his data?
- Which hypothesis tests, and why, should Mr. Brown employ to evaluate the success of the vocabulary acquisition strategy?
- What limits should Mr. Brown be aware of when extrapolating his conclusions from the outcomes of the hypothesis tests?
A dataset of vocabulary scores before and after the learning technique was implemented in Mr. Brown’s classrooms should be given to the students. In order to do the data analysis, they need also have access to SPSS Statistics software.
TYPES OF DATA
Researchers may collect various types of numeric data when conducting a quantitative study in language teaching. The type of data collected can influence the research design and data analysis. There are four distinct types of numeric data: nominal, ordinal, interval, and ratio. It is essential to understand the unique properties of each type of data to select the appropriate analysis method and draw valid conclusions.
Nominal data is divided up into many groups or categories without a natural hierarchy or order. Nominal information may be used to indicate traits like gender, ethnicity, or native language. Nominal data may be analyzed using descriptive statistics, including frequency distributions and contingency tables (Kottler & Keller, 2009).
Data that is ranked or ordered according to some criteria is referred to as ordinal data. However, the distance between each rank is unknown. Language competence levels, for instance, might be categorized as beginning, intermediate, or advanced. Although the separation between each level is unknown, they may be arranged. Ordinal data may be analyzed using descriptive statistics like median and mode (Gall et al., 2007).
Data having an ordered scale and equal gaps between the values is referred to as interval data since it lacks a real zero point. Temperature and dates on a calendar are two examples of interval data. Descriptive statistics like mean and standard deviation may be used to examine interval data (Field, 2013).
Data having an ordered scale, equal gaps between the numbers, and a real zero point are referred to as ratio data. The ratio data types include time, weight, and height. Descriptive statistics like mean, standard deviation, and range may be used to examine ratio data (Salkind, 2010).
When doing quantitative research, choosing the best study design and data analysis techniques requires understanding the many forms of numerical data. When evaluating data for language instruction research, nominal, ordinal, interval, and ratio data characteristics should be considered.
Watch this video clip about Nominal, Ordinal, Interval & Ratio Data
NOMINAL DATA
Nominal data is a subset of categorical data that consists of categories or values that are not numeric and do not naturally have a natural order or ranking. Nominal data is often used in language instruction research to classify participants according to demographic factors such gender, age, native language, and degree of language ability.
For instance, Dornyei and Ottó (1998) obtained nominal data on the participants’ gender by asking them to self-identify as male or female. This study examined the connection between gender and motivation in language acquisition. Similar to this, the researchers in a study (Bitchener & Knoch, 2009) that looked at the impact of various forms of feedback on writing ability gathered nominal data on the participants’ original languages by classifying them as either native speakers or non-native speakers of English.
Language competency level is another example of a nominal variable in research on language instruction. Researchers often classify participants into competence levels like A1, A2, B1, B2, C1, and C2 using measures like the Common European Framework of Reference for Languages (CEFR) or the Test of English as a Foreign Language (TOEFL). For instance, the researchers used the CEFR scale to gather nominal data on the participants’ language competence levels in a study that examined the impact of learner autonomy on language competency.
Frequency distributions, contingency tables, and chi-square tests are often used in the analysis of nominal data to evaluate the connections between categories and spot data patterns and trends.
The following table is an example for nominal data:
| Participant | Gender | Age Group | Native Language | Language Proficiency Level |
| 1 | Female | 18-24 | Spanish | B1 |
| 2 | Male | 18-24 | Korean | A2 |
| 3 | Female | 25-34 | Arabic | C1 |
| 4 | Male | 35-44 | French | B2 |
| 5 | Female | 45-54 | German | A1 |
| 6 | Male | 18-24 | Chinese | B1 |
| 7 | Female | 25-34 | Japanese | C2 |
| 8 | Male | 35-44 | Russian | A2 |
| 9 | Female | 45-54 | Italian | B2 |
| 10 | Male | 18-24 | Portuguese | A1 |
The gender, age group, native language, and degree of linguistic skill are all included in this example’s nominal data. The data is divided into many groups or categories, but there is no intrinsic hierarchy or ranking within the categories. The data can be analyzed using frequency distributions and contingency tables to identify patterns and relationships between the categories.
Input the nominal data table in SPSS:
- Open SPSS and click on “Type in Data” option.
- Enter the variable names in the first row of the data editor, which includes “Participant”, “Gender”, “Age Group”, “Native Language”, and “Language Proficiency Level”.
- Enter the corresponding data under each variable name. For example, under the “Gender” variable, enter “Female” or “Male” for each participant. Similarly, enter the appropriate data under each of the other variables.
- Ensure that the data type for each variable is set to “string” by clicking on the variable name in the data editor and selecting “string” under “Variable View.”
- Save the data as a SPSS file by clicking on “File” and selecting “Save As.”
- Name the file, select the appropriate file type (e.g. .sav), and click “Save.”
After entering the data into SPSS, you can run descriptive statistics such as frequency distributions and contingency tables to analyze the nominal data.
ORDINAL DATA
Ordinal data is a type in which the categories can be ranked or ordered based on some criterion, but the distance between each rank is unknown. For instance, students can be categorized into beginner, intermediate, and advanced language proficiency levels. The ranking indicates that advanced proficiency level students have more skills than intermediate level students, but the distance between each level is unknown. In language teaching research, ordinal data can be used to represent variables such as language proficiency level, academic achievement, or teacher effectiveness.
Ordinal data are often analyzed using descriptive statistics in studies on language training. Because they are unaffected by extreme values, in contrast to the mean, the median and mode are the most often employed measures of central tendency in ordinal data analysis. The mode is the value that occurs the most often in the data collection, while the median is the number that divides the data into two equally sized half (Gall et al., 2007).
Percentile rankings may be used to evaluate ordinal data in addition to the median and mode. Percentile rankings show the proportion of a group’s members who scored below a certain mark. Percentile rankings may be used to assess how a student’s score compares to other students’ scores or to find kids who would benefit from further instruction (Kline, 2016).
Research on language education benefits from the use of ordinal data since it enables us to rank or arrange significant factors. Ordinal data analysis often makes use of descriptive statistics like median, mode, and percentile rankings.
This is an example for ordinal data with n = 10
| Participant | Language proficiency level |
| 1 | Beginner |
| 2 | Intermediate |
| 3 | Advanced |
| 4 | Beginner |
| 5 | Intermediate |
| 6 | Intermediate |
| 7 | Beginner |
| 8 | Advanced |
| 9 | Intermediate |
| 10 | Intermediate |
The variable of relevance in this case is the language skill level, which has three levels: beginner, intermediate, and advanced. The participants are rated or arranged according to their degree of language skill, although it is unknown how far off each level is from the others. As a result, this data may be regarded as ordinal data.
Into SPSS, enter the ordinal data:
- Launch SPSS and choose “File” from the top-left menu.
- Click “New” and then “Data.”
- Select “Numeric” as the “Type” and type “Participant” into the first column of the “Name” field in the “Variable View” tab.
- Set the “Type” to “Ordinal” and input “Language proficiency level” in the second column’s “Name” box.
- Double-click on each cell to enter the numbers for each participant in the “Language proficiency level” column. Then, choose the proper level (Beginner, Intermediate, or Advanced) from the drop-down option.
- When all of the values have been input, click “File” in the upper left corner and choose “Save” to save the data file.
- Click “Analyze” in the top menu, choose “Descriptive Statistics,” then “Frequencies,” to run descriptive statistics on this collection of data.
- Select the variable “Language proficiency level” in the “Frequencies” dialog box, then click “OK” to start the analysis.
- Descriptive statistics including the number of participants in each level, the percentage of participants in each level, and the data’s mode will be given to you by SPSS. The median and other metrics of central tendency for this data set may also be determined using SPSS’s “Explore” feature.
INTERVAL DATA
Data having an ordered scale and equal gaps between the values is referred to as interval data since it lacks a real zero point. Temperature and dates on a calendar are two examples of interval data. Descriptive statistics like mean and standard deviation may be used to examine interval data (Field, 2013).
One of the four categories of quantitative or continuous data is interval data. It is distinguished by having a scale that is ordered with equal gaps between the numbers but no actual 0 point. Test scores are a typical illustration of interval data since they might fall anywhere on a number line within the bounds of a specified data collection. A test score range, for instance, may be from 0 to 100; a student might get any score within this range. Using descriptive statistics like mean, standard deviation, and range, interval data may be quickly recorded and examined in SPSS.
Research on language training may also make use of interval data. For instance, the results of the students may be gathered as interval data in a research that tries to examine the impact of a particular teaching strategy on students’ performance on a language proficiency test. There is no genuine zero point in this situation, but the scores have an ordered scale with equal distances between the numbers. The difference between the two scores that are next to each other is the same, but a score of zero does not equal total ignorance.
It’s crucial to understand that the variations in the values are absolute rather than relative. As an example, the difference between a grade of 5 and a grade of 6 is equivalent to 1 point, or the difference between a grade of 1 and a grade of 2. However, we cannot claim that a person with a score of 6 is three times more knowledgeable than a person with a score of 2. Instead, we can only state that the person who had a score of 6 properly answered 60% of the questions and the one who received a score of 2 correctly answered 20%.
The interval scale cannot be used to demonstrate complete mastery of a topic, either. Scores solely represent a thorough understanding of or ignorance of the test questions. Other examples of interval-level data include temperature, aptitude scores, and intelligence quotients. Therefore, it is important to use appropriate statistical tests and methods when analyzing interval data in language teaching research.
Descriptive statistics such as mean and standard deviation can be used to analyze interval data in language teaching research. The mean provides an estimate of the central tendency of the data, while the standard deviation measures the variability of the data around the mean. Additionally, inferential statistics such as t-tests or ANOVA can be used to compare groups or test hypotheses about the data.
An example of interval data:
Suppose a researcher wants to investigate the effectiveness of a language intervention program on the speaking proficiency of 10 ESL learners. The speaking proficiency scores of the learners before and after the intervention are collected on a scale of 0-100, with higher scores indicating higher proficiency. The collected data can be presented in the following table:
| Learner | Before Intervention | After Intervention |
| 1 | 50 | 60 |
| 2 | 75 | 80 |
| 3 | 60 | 70 |
| 4 | 70 | 75 |
| 5 | 45 | 55 |
| 6 | 65 | 70 |
| 7 | 80 | 90 |
| 8 | 55 | 60 |
| 9 | 70 | 75 |
| 10 | 30 | 45 |
In this case, the scores have an ordered scale with equal intervals between the values, but there is no true zero point. The difference between the two adjacent scores is the same, but a score of zero does not indicate a complete lack of knowledge.
RATIO DATA
Research on language training may make use of ratio data, which is a kind of quantitative or continuous data. For instance, a ratio scale may be used to quantify each participant’s word acquisition in a research examining the efficacy of a language learning program. A participant who learnt 0 words did not learn any words at all. The quantity of words learned would have an absolute zero point. Additionally, a real ratio might be used to describe the variation in the amount of words learnt across individuals, enabling more accurate data analysis.
Data with a real zero point and values on a scale with equal gaps between them are referred to as ratio data. This indicates that mathematical operations like addition, subtraction, multiplication, and division may be used to compare the values. The number of vocabulary items learnt, the amount of time needed to complete a task, or the quantity of mistakes in a writing sample are a few examples of ratio data in language education research (Salkind, 2010).
The amount of accurate responses on a language competence test might serve as another example of ratio data in language education research. Because a person who responded 0 properly did not answer any questions at all, this data may also be quantified using a ratio scale, and the difference in the proportion of right responses across participants can be stated as a true ratio.
Ratio data may be analyzed using descriptive statistics like mean, standard deviation, and range. The data’s mean is determined by adding together all the values and dividing by the total number of values. How much the data depart from the mean is shown by the standard deviation. The range is the discrepancy between the data set’s highest and lowest values.
For instance, in a study on language competence, researchers may gather information on the quantity of vocabulary terms that a set of students had mastered. If the researchers have ratio data on this variable, they could use descriptive statistics to calculate the data’s mean, standard deviation, and range. These statistics could then be used to draw conclusions about the participants’ vocabulary acquisition and to compare it with other groups.
An example of a data table for ratio data:
| Participant | Age | Speaking Fluency Score | Writing Fluency Score |
| 1 | 23 | 82 | 89 |
| 2 | 25 | 91 | 87 |
| 3 | 29 | 76 | 82 |
| 4 | 31 | 94 | 91 |
| 5 | 27 | 80 | 84 |
| 6 | 24 | 87 | 92 |
| 7 | 28 | 93 | 89 |
| 8 | 26 | 85 | 83 |
| 9 | 30 | 88 | 91 |
| 10 | 32 | 89 | 85 |
In this example, age, speaking fluency score, and writing fluency score are all examples of ratio data as they have a true zero point and equal intervals between values.
Summary table for types of data:
| Type of Data | Description | Examples in Language Education |
| Nominal | Categorical data that are assigned labels, but no order or numerical values | Nationality, gender, language proficiency level |
| Ordinal | Categorical data are assigned labels with a specific order or ranking, but the difference between values is not constant | Language proficiency levels (beginner, intermediate, advanced), Likert scale ratings |
| Interval | Numerical data where the distance between two points is equal, but there is no true zero point | Temperature, time, scores on a standardized test |
| Ratio | Numerical data where there is a true zero point, allowing for meaningful ratios to be calculated | Number of words spoken or written, number of correct answers on a test |
The country, gender, and degree of language competency of language learners are only a few examples of nominal data in language education. Language competence levels, such as beginner, intermediate, and advanced, as well as Likert scale evaluations, which are often used to gauge students’ attitudes about various facets of language acquisition, are examples of ordinary data in the field of language education. The temperature in the classroom, time spent learning or using the language, and results on standardized language examinations are all examples of interval data in language education. The quantity of words that students speak or write, the quantity of right answers on a test, or the quantity of time spent studying are all examples of ratio data in language education.
TEST THE HYPOTHESES
A statistical method known as hypothesis testing is used in research to ascertain if the results are consistent with a hypothesis on a population parameter. Hypothesis testing may be used to assess the efficacy of language education strategies, resources, and approaches.
The null and alternative hypotheses must be stated before conducting a hypothesis test. The alternative hypothesis (Ha) is the hypothesis that there is a significant difference, whereas the null hypothesis (H0) is the hypothesis that there is no significant difference between the groups being compared. The null hypothesis would be that there is no significant difference between the effectiveness of the two methods, while the alternative hypothesis would be that there is a significant difference, for instance, if a researcher wants to test the hypothesis that one method of teaching a language is more effective than another method.
The choice of an appropriate statistical test for the data analysis is the next stage. The test to use relies on the research question being addressed as well as the kind of data being examined (nominal, ordinal, interval, or ratio). The chi-square test would be the proper test, for instance, if the data is nominal and the study question is if there is a significant difference in language competency between two groups of learners. The t-test would be the proper test if the data were interval or ratio data, and the study question was “Is there a significant difference in language proficiency between two groups of learners?”
The researcher gathers data, computes the test statistic, and determines the p-value after choosing the right statistical test. The p-value shows the likelihood of finding the observed data or more extreme data if the null hypothesis is true, while the test statistic evaluates the difference between the observed data and the null hypothesis. The alternative hypothesis is accepted and the null hypothesis is rejected if the p-value is less than the significance threshold, which is typically set at 0.05. The null hypothesis is not disproved if the p-value exceeds the significance threshold.
By analyzing data to test a hypothesis and establish whether it is supported or not, statistical analysis is one method of doing hypothesis testing in language instruction. Setting up the null and alternative hypotheses, choosing the best statistical test, computing the test statistic, and calculating the p-value are all phases in this process (Mackey & Gass, 2021).
Hypothesis testing can be used in language teaching research to explore a range of issues, including whether one teaching approach is more effective than another, whether language proficiency and age of acquisition are related, and whether a particular instructional intervention has an effect on students’ attitudes toward language learning (Larsen-Freeman & Anderson, 2013).
A research by Larsen-Freeman and Long (2014) that evaluated the efficacy of three distinct forms of corrective feedback on students’ use of the English article system serves as an example of hypothesis testing in language instruction. The alternative hypothesis, that there would be a substantial difference between the three forms of feedback, was put up as the alternative to the null hypothesis that there would be no significant difference between the three types of feedback. The data was then examined using an ANOVA test, and it was discovered that one kind of feedback was much more efficient than the other two.
In short, hypothesis testing is useful in language education research because it enables academics to put their hypotheses and concepts to the test using real-world evidence. It ensures that instructional strategies and interventions are supported by research and are successful, which benefits both educators and language learners in the long run.
NULL HYPOTHESIS
A statement that asserts there is no significant link between two variables or distinction between groups under comparison is known as a null hypothesis in statistics. It serves as the starting point for statistical tests, the goal of which is to reject or not reject the null hypothesis. Null hypotheses in language education may be used to examine the efficacy of language learning strategies or the association between language competency and certain variables (McDonough & McDonough, 2014).
A null hypothesis in language instruction can state, for instance, that there is no appreciable difference between pupils who learnt English in a conventional classroom and those who did so online. The null hypothesis presupposes that there is no discernible difference in the two groups’ levels of linguistic competence. After gathering information from both groups, a researcher might use statistical tests to determine whether or not to reject the null hypothesis. The alternative hypothesis that there is a substantial difference between the two groups may be supported if the null hypothesis is rejected and there is evidence of a significant difference in language proficiency between the two groups.
Another example of a null hypothesis in language instruction is that there is no connection between a student’s degree of motivation and their language competency. Once again, the null hypothesis makes the assumption that there is no meaningful connection between the two variables. The researcher might then gather information on motivation levels and degrees of language competency and use statistical tests to examine it. If the alternative hypothesis—that there is a substantial relationship—is supported, it suggests that there is evidence of a significant association between motivation level and language proficiency.
In order to determine if there is a statistically significant difference between two groups or circumstances, null hypothesis testing is utilized in the area of language instruction. The idea that there is no discernible difference between the groups or circumstances under study is known as the null hypothesis.
For instance, a language instructor could be curious to find out if a certain teaching strategy affects pupils’ writing skills. The null hypothesis would be that there is no discernible difference between the writing abilities of pupils using the new teaching approach and those using the conventional one in this situation.
The instructor would undertake a research in which two groups of pupils were randomly allocated to either the new or conventional methods of instruction in order to test this null hypothesis. After some time has passed, statistical analysis is used to gauge and compare the writing prowess of the two groups. If the findings indicate that there is no substantial difference in the two groups’ levels of writing skill, the null hypothesis would be validated.
To ascertain whether an intervention or therapy has a statistically significant impact on language learning outcomes, null hypothesis testing is often utilized in language education research. It enables scientists to draw unbiased conclusions from empirical data and ensures that any observed variations between groups are not the result of random chance.
The efficacy of language learning techniques, the association between language competency and certain variables, and other research topics in language education may all be tested using null hypotheses as the starting point for statistical studies.
ALTERNATIVE HYPOTHESIS
The alternate hypothesis in a hypothesis test is the exact opposite of the null hypothesis. It is a claim that there is a connection between the variables or a distinction between the groupings. It is, in other words, the hypothesis that the researcher is attempting to validate or support. According to Campbell and Stanley (1963), the alternative hypothesis might be either directed, specifying the impact’s direction, or non-directional, only claiming that an effect is there.
The alternative hypothesis may be used to language instruction to evaluate the efficacy of a novel technique or teaching strategy. An alternate theory would state, for instance, that students who are taught using a communicative approach to language learning will do better on a speaking examination than students who are taught using a grammar-translation strategy. The alternate theory in this situation is that there is a considerable discrepancy between the two student groups’ speaking exam results.
A research would need to gather information from both groups of students and use statistical techniques to evaluate the findings in order to test the alternative hypothesis. If the data reveal a substantial difference in speaking test scores between the two groups, the alternative hypothesis would be validated.
SPSS STATISTICS
One of the many industries that uses the potent software package SPSS Statistics for statistical analysis is the teaching of languages. Data management, manipulation, and analysis are made possible, and charts and graphs may be made to graphically depict the data. The descriptive statistics mean, median, mode, standard deviation, and frequency distributions may all be explored by language instructors and researchers using SPSS Statistics. Inferential statistical tests can also be run to look at links between variables, differences between groups, and other things. A crucial tool for any language instructor or researcher looking to comprehend their data and reach meaningful conclusions is this program, which is extensively used in language research due to its capacity for handling enormous datasets and intricate analysis.
FREQUENCY DATA
The frequency of a certain value or category inside a dataset is referred to as frequency data. It is a kind of categorical data, sometimes shown as a frequency distribution that shows the frequency of each value or category within a dataset.
Frequency statistics may be used in language instruction to examine how a language is utilized in a given situation or by a particular set of speakers. For instance, a language instructor may compile information on how often students use certain words in essays or class discussions. This information may then be used to influence instructional choices by being evaluated to spot frequent mistakes or areas of a student’s language skills that need improvement.
In a wider context, such as a corpus of written or spoken language, frequency statistics may also be used to assess language usage. To shed light on linguistic use trends in a certain genre or social setting, a corpus linguist can, for instance, examine the prevalence of particular grammatical constructions or lexical items in a variety of texts or voice samples.
For example, measures of central tendency (such as mean, median, and mode) and measures of dispersion (such as standard deviation, range) may be used to assess frequency data in language instruction. These measurements may assist find trends or outliers by revealing important details about the dataset’s distribution of values or categories.
In general, frequency data is a useful resource for studying language usage and may reveal patterns and trends in language understanding and output.
This is an example of a frequency data table with n = 10:
| Language | Gender | Age | Country | Frequency |
| English | Male | 22 | USA | 3 |
| French | Female | 25 | Canada | 2 |
| Spanish | Male | 30 | Mexico | 1 |
| German | Female | 28 | Germany | 1 |
| English | Female | 24 | UK | 2 |
| Arabic | Male | 20 | Egypt | 1 |
The table in this illustration displays the frequency of the various languages used by 10 language learners. It contains details on their age, gender, and country of origin. The frequency column shows the number of language learners for each language.
DESCRIPTIVE STATISTICS
In order to characterize and summarize data, descriptive statistics are crucial tools in data analysis and are often employed in the area of teaching languages. They may be used to explain a collection of data’s central tendency, variability, and dispersion. Descriptive statistics, according to Gravetter and Wallnau (2014), aid in the organization and distillation of enormous volumes of data into formats that are easier to handle and comprehend.
A popular piece of software for doing statistical studies, including descriptive statistics, is SPSS (Statistical Package for the Social Sciences). Researchers may utilize SPSS to compute metrics like the mean, median, and mode to characterize a data set’s central tendency in language education. To characterize the variability of the data, they may also compute metrics like the range, standard deviation, and variance.
Researchers may enter their data into the program and choose the relevant statistical measures they want to compute in order to examine descriptive statistics in SPSS. They may choose the “Frequencies” option, for instance, to learn more about the frequency distribution of their data, or the “Descriptives” option to learn more about the variability and central tendency of their data.
In order to characterize and summarize data, descriptive statistics are helpful tools in language instruction research. For researchers to undertake numerous descriptive statistical studies, SPSS offers an intuitive platform.
This is an example for mean (M), and standard deviation (SD):
| Participant | Test Score |
| 1 | 85 |
| 2 | 90 |
| 3 | 78 |
| 4 | 80 |
| 5 | 92 |
| 6 | 88 |
| 7 | 84 |
| 8 | 87 |
| 9 | 91 |
| 10 | 86 |
Mean (M) = (85 + 90 + 78 + 80 + 92 + 88 + 84 + 87 + 91 + 86) / 10 = 86.1
Standard deviation (SD) = 4.87
In this instance, the 10 participants’ mean test scores were 86.1, with a standard deviation of 4.87.
View this video on YouTube.
Descriptive Statistics and z Scores in SPSS – SPSS for Beginners
PRACTICE IN SPSS
An example of a data table with descriptive data with n = 20. Run the descriptive statistics for ages, genders, and years after entering the data into SPSS.
A sample data
| Participant | Age | Gender | Nationality | Years of English Learning |
| 1 | 23 | Male | Japanese | 10 |
| 2 | 29 | Female | Korean | 5 |
| 3 | 27 | Male | Brazilian | 8 |
| 4 | 31 | Male | French | 2 |
| 5 | 25 | Female | Chinese | 7 |
| 6 | 30 | Male | Russian | 3 |
| 7 | 28 | Female | Spanish | 9 |
| 8 | 26 | Male | Turkish | 6 |
| 9 | 24 | Female | Indian | 11 |
| 10 | 32 | Male | German | 4 |
STATISTICAL SIGNIFICANCE WITH THE T-TEST
The statistical significance test determines whether differences between groups or variables are more likely to be the result of chance than to reflect actual differences. The t-test is a frequently used statistical test to evaluate the statistical significance of differences between two groups or variables in language instruction research.
To compare the means of two groups or variables, use the t-test. It presupposes that the data are regularly distributed and may be used to interval or ratio data. To evaluate if there is a statistically significant difference between two groups or variables, the t-test computes a t-value and compares it to a critical value. If the t-value exceeds the critical threshold, the difference is deemed statistically significant.
The independent samples t-test and the paired samples t-test are the two different kinds of t-tests. The means of two distinct groups are compared using the independent samples t-test, whereas the means of two related groups are compared using the paired samples t-test.
The researcher must first create a null hypothesis, which states that there is no significant difference between the two groups or variables being compared, before they can perform a t-test. According to the alternative theory, the two groups or variables vary significantly from one another. When the threshold of significance, or alpha level, is set at.05, there is a 5% probability that the null hypothesis will be rejected even if it is true.
Let’s say the t-test produces a result that is statistically significant. In such situation, it indicates that there is a substantial difference between the two groups or variables and that the null hypothesis may be rejected. This may provide insightful information on the efficiency of language teaching techniques or the effects of language learning on various factors.
To put it briefly, statistical significance using the t-test is a crucial tool for deciphering and evaluating data in studies on language training. Researchers may decide on the efficacy of various teaching techniques and the influence of language acquisition on various variables by comparing means and evaluating statistical significance.
THE INDEPENDENT SAMPLES T-TEST
A statistical test called the independent samples t-test is used to evaluate if there is a significant difference between the means of two independent groups. This exam may be used in language instruction to compare how two groups of students performed on a certain language task or ability.
The data from each group must be regularly distributed and have equal variances in order to perform an independent samples t-test. The alternative hypothesis is that there is a significant difference, while the null hypothesis for the test is that there is no significant difference between the means of the two groups.
By dividing the difference between the means of the two groups by the standard error of the difference, the t-test generates a t-value. Then, using a significance threshold determined by the researcher, the t-value is compared to a critical value taken from a t-distribution table with degrees of freedom equal to the difference between the two groups’ combined sample sizes, minus two.
Assume that the estimated t-value exceeds the crucial level. In such situation, it is determined that there is a substantial difference between the means of the two groups and that the null hypothesis is not accepted. The null hypothesis, on the other hand, cannot be ruled out if the estimated t-value is smaller than the critical value, leading to the conclusion that there is no appreciable difference between the means of the two groups.
The independent samples t-test is a valuable tool for comparing the efficacy of various teaching strategies, the performance of students from varied linguistic backgrounds, or the effects of various instructional interventions on language learning outcomes in research on language teaching.
One example of applying the independent samples t-test in language teaching is a study by Pham (2022) which aimed to investigate whether engaging in collaborative writing activities would affect the quality of writing of each student in an academic writing course for argumentative essays. The study involved 62 third-year English majors from the Faculty of Foreign Languages at Van Lang University in Ho Chi Minh City, Vietnam, with 35 students in the experimental group and 27 students in the control group, aged between 19 to 21. The teaching and learning activities for both groups were similar except for the essay-composing stage, where the control group composed their essays individually, while the experimental group composed their essays collaboratively. The study analyzed data from pre- and post-tests to compare the writing quality of both groups. The independent sample t-tests showed that collaborative writing activities significantly improved each student’s writing quality, as compared to individual writing.
Watch this video clip on YouTube about Independent Samples t-Tests in SPSS – SPSS for Beginners
PRACTICE IN SPSS
Enter the information from the following table into SPSS, then use the Independent sample t-test to see if the test results between the two groups vary.
Control Group
| Participant | Pre-Test Score | Post-Test Score |
| 1 | 55 | 73 |
| 2 | 75 | 80 |
| 3 | 60 | 68 |
| 4 | 63 | 70 |
| 5 | 67 | 77 |
| 6 | 57 | 65 |
| 7 | 67 | 72 |
| 8 | 56 | 66 |
| 9 | 60 | 73 |
| 10 | 72 | 81 |
Experimental group
| Participants | Pre-test score | Post-test score |
| 1 | 60 | 82 |
| 2 | 70 | 85 |
| 3 | 55 | 78 |
| 4 | 65 | 76 |
| 5 | 75 | 90 |
| 6 | 50 | 74 |
| 7 | 65 | 80 |
| 8 | 55 | 75 |
| 9 | 60 | 83 |
| 10 | 70 | 85 |
THE PAIRED SAMPLES T-TEST
A statistical test called the paired samples t-test is used to detect if there is a significant difference between two sets of related data. This test may be used in language education to establish if there is a substantial change in a group of students’ language proficiency before and after a particular intervention, such a language course or tutoring program.
The pre- and post-test scores of a set of students are an example of paired data that may be compared using the paired samples t-test. The test assesses if there is a statistically significant difference between the two sets of data and measures that difference.
The data must be regularly distributed and the data pairs must be independent in order to execute the paired samples t-test. The test generates a t-statistic, which is used to assess the significance of a difference between two sets of data or whether it might have happened by chance.
When the paired samples t-test yields a significant result, it means that there is a substantial difference between the two sets of data, such as the language proficiency of the students before and after an intervention. This might be helpful when assessing a language program’s or a teacher’s method’s efficacy.
A research that assesses the efficacy of a novel method to language instruction is one instance of a paired samples t-test in language education. Before the intervention, a set of students might take a pre-test to gauge their level of language competence, and after the intervention, they could take a post-test. If there is a significant difference in language competence between the pre-test and post-test scores, the paired samples t-test might be employed to detect this.
In general, the paired samples t-test is a practical statistical instrument for assessing the efficacy of language education methods and programs. It enables pre- and post-test score comparison and provides light on the efficiency of the intervention.
Watch this video clip on YouTube about Paired Samples t-Tests in SPSS – SPSS for Beginners
PRACTICE IN SPSS
Enter the information into SPSS and use the paired sample t-test to compare the students’ mean scores between their pre- and post-tests.
Table 4. pre-test and post-test
| Participant | Pre-Test Score | Post-Test Score |
| 1 | 15 | 20 |
| 2 | 18 | 22 |
| 3 | 12 | 16 |
| 4 | 20 | 24 |
| 5 | 16 | 18 |
| 6 | 13 | 19 |
| 7 | 17 | 21 |
| 8 | 14 | 17 |
| 9 | 19 | 23 |
| 10 | 11 | 15 |
ONE-SAMPLE T-TESTS IN SPSS
A statistical analysis technique called a one-sample t-test is used to compare the average score of a sample to the average of the known population or an estimated value. It may be used in language instruction to determine if a set of language learners’ mean scores on a particular activity or language competence diverge considerably from the population mean or a predetermined value.
The sample data and the population mean or estimated value are required for the one-sample t-test in SPSS. Researchers may use the “Analyze” > “Compare Means” > “One-Sample T Test” menu options after inputting the data into SPSS. The population mean or predicted value may then be entered in the “Test Value” box once the researcher selects the variable of interest.
The sample’s mean and standard deviation, the test result, the t-value, the degrees of freedom, and the p-value are all included in the output of a one-sample t-test in SPSS. The significance of the finding is assessed using the t-value, with higher t-values indicating a wider discrepancy between the sample mean and the predicted value. The p-value provides information on the likelihood that the observed result could have come about by chance and is used to assess whether the result is statistically significant or not.
Investigating if a group of language learners’ mean score on a vocabulary test differs substantially from the population mean of vocabulary scores for their age group is one instance of employing a one-sample t-test in language education. Researchers may evaluate if a group’s performance is above or below the typical performance level for their age group by using a one-sample t-test.
Watch this video clip on YouTube about One-Sample t-Tests in SPSS – SPSS for Beginners.
PRACTICE IN SPSS
Utilizing the information from the pre- and post-test data in the table of the Paired Sample T-Tests, compare the post-test mean scores to the population mean score of 22 (Population M = 22), which represents the typical mean score for all courses.
CORRELATION IN SPSS
A statistical method for determining the link between two or more variables is correlation. Correlation may be used in language teaching to investigate the connections between language competency and other characteristics like age, gender, educational attainment, or language learning methods.
The Correlations technique in SPSS may be used to perform correlation. This process determines the significance levels of the correlation coefficients between pairs of variables. A correlation matrix displaying the pairwise correlations between the relevant variables is included in the Correlations procedure’s output.
The Pearson’s r correlation coefficient, which gauges the linear connection between two continuous variables, is the most often used correlation coefficient. Nevertheless, several correlation coefficients may be used based on the variables’ characteristics, such as Spearman’s rho for ordinal variables or point-biserial correlation for a continuous and a binary variable.
Examining the direction and intensity of the link is necessary for interpreting correlation coefficients. The range of the correlation coefficient is 0 to 1, with 1 denoting a perfect negative correlation and -1 denoting a perfect positive correlation. The magnitude of the correlation coefficient reflects the strength of the link; bigger absolute values signify stronger associations. Positive values of the correlation coefficient indicate a positive association, whereas negative values suggest a negative relationship, which is shown by the sign of the coefficient.
For instance, a correlation study in language instruction may show that language competency and the frequency of language usage outside of the classroom have a substantial positive association (r =.70, p .05). This would imply that pupils tend to have better levels of competency when they utilize the language more often outside of the classroom.
For examining the connections between variables in language education research, correlation analysis is a useful tool, and SPSS offers a user-friendly interface for carrying out these studies.
Watch this video clip on YouTube about Correlation in SPSS – SPSS for Beginners
PRACTICE IN SPSS
An example data table including age, gender, education level, and language learning techniques is shown in the table below. To run the correlation, enter the information from the following tables into SPSS.
| Participant | Age | Gender | Education Level | Strategy 1 | Strategy 2 | Strategy 3 |
| 1 | 25 | Female | Bachelor’s | 5 | 4 | 2 |
| 2 | 32 | Male | Master’s | 3 | 5 | 4 |
| 3 | 19 | Female | High School | 2 | 2 | 3 |
| 4 | 28 | Male | PhD | 4 | 3 | 5 |
| 5 | 21 | Female | Bachelor’s | 3 | 4 | 2 |
| 6 | 30 | Male | Master’s | 4 | 3 | 4 |
| 7 | 18 | Female | High School | 2 | 2 | 2 |
| 8 | 27 | Male | Bachelor’s | 5 | 4 | 3 |
| 9 | 23 | Female | Master’s | 4 | 5 | 4 |
| 10 | 26 | Male | PhD | 3 | 3 | 5 |
In this illustration, the table shows participant personal data, including age, gender, educational attainment, and their reported usage of three distinct language-learning methodologies (Strategy 1, Strategy 2, and Strategy 3). The information might be used to investigate connections between demographic variables and language learning techniques or to look at variations in approach utilization amongst participant groups according to age, gender, or educational attainment.
Note: The Correlation is also used to test if the inter-raters’ scores are correlated.
SIMPLE LINEAR REGRESSION IN SPSS
By fitting a linear equation to the observed data, simple linear regression is a statistical technique that enables us to examine the connection between two continuous variables. Simple linear regression is often used in language education research to examine the relationship between two variables, such as the connection between language ability and exposure to the language.
The dependent variable (also known as the criterion variable) and the independent variable (also known as the predictor variable) must first be specified in order to perform a basic linear regression analysis in SPSS. Following the identification of these variables, SPSS will compute the regression equation, which enables us to forecast the value of the dependent variable depending on the value of the independent variable.
Simple linear regression analysis in SPSS produces a number of tables, including an ANOVA table, a coefficients table, and a description of the model. The R-squared value, which is a measure of how much variation in the dependent variable is explained by the independent variable, is one piece of information regarding the model’s overall fit that is included in the summary of the model table. The regression line’s slope (beta coefficient), intercept, standard error, and degree of significance are all detailed in the coefficients table. The overall significance of the regression model is shown in the ANOVA table.
It is crucial to remember that basic linear regression presumes a linear connection between the dependent and independent variables and that the residuals (the discrepancies between the actual values and those predicted) are normally distributed and have constant variance.
Watch this video clip on YouTube about How to do Simple Linear Regression in SPSS
PRACTICE IN SPSS
To do a simple linear regression in SPSS, enter an example data table containing learner autonomy, foreign language anxiety, and language competency.
| Participant | Learner Autonomy (LA) | Foreign Language Anxiety (FLA) | Language Proficiency (LP) |
| 1 | 4.2 | 2.5 | 3.8 |
| 2 | 3.7 | 3.9 | 2.6 |
| 3 | 4.1 | 2.8 | 4.2 |
| 4 | 2.6 | 4.3 | 1.8 |
| 5 | 3.9 | 3.1 | 3.4 |
| 6 | 4.4 | 2.0 | 4.9 |
| 7 | 3.1 | 3.5 | 2.9 |
| 8 | 3.8 | 2.7 | 3.7 |
| 9 | 4.3 | 3.8 | 4.1 |
| 10 | 2.9 | 4.2 | 2.2 |
This table shows the results of three separate measurements: learner autonomy (LA), foreign language anxiety (FLA), and language proficiency (LP). Each row represents a different research participant. greater values indicate greater levels of the related construct, while the numbers in each column reflect the participant’s score on that measure. Through the use of a correlation or regression analysis, this table might be utilized to investigate the connections between these three variables.
Open SPSS and choose the “Analyze” tab as the first step.
Step 2: From the list of choices, choose “Regression”.
Step 3: From the list of regression possibilities, choose “Linear”.
Step 4: From the list of variables, choose the dependent variable. The variable you wish to forecast is this one.
Step 5: From the list of variables, choose the independent variable or variables. You will utilize these variables to forecast the dependent variable.
Step 6: To perform the regression, click “OK”.
Step 7: Go through the results. The regression equation, the coefficient of determination (R2), and the p-value are all included in the output.
Step 8: Analyze the outcomes. The strength of the link between the independent and dependent variables is shown by the coefficient of determination (R2). The likelihood that the link is the result of chance is indicated by the p-value.
You may simply conduct a simple linear regression in SPSS by following these instructions. To make meaningful inferences from the regression findings, it’s crucial to remember to interpret the data.
ONE-WAY ANOVA
One-way In order to compare the means of three or more independent groups, language instruction research often use the statistical approach known as ANOVA (Analysis of Variance). The fundamental concept behind an ANOVA is to divide a variable’s total variance into two parts: the variance between groups and the variance within groups. While the variance within groups shows the variety within each group, the variance between groups represents the variations in group means.
One-way ANOVA can be used to find the answers to research questions like whether there are differences in reading comprehension scores between students who have received various types of instruction or whether there are differences in speaking proficiency levels between students at various proficiency levels.
The stages involved in doing a one-way ANOVA are as follows:
- Identify the study’s null hypothesis (H0) and alternative hypothesis (Ha).
- Gather information from the comparison groups.
- Determine the sample size, variance, and mean for each group.
- Determine the total sum of squares (SST), which is the total of all the observations’ squared deviations from the mean.
- Determine the sum of squares between (SSB), which is the sum of the squared differences between group means and the overall mean, weighted by the sample size.
- Determine the sum of squares within (SSW), or the sum of the squared deviations of each observation from the mean of the group.
- Determine the SSB and SSW degrees of freedom (df).
- By dividing SSB and SSW by the corresponding degrees of freedom, calculate the mean square (MSB) and mean square within (MSW).
- Dividing the MSB by the MSW yields the F-statistic.
- Use a table of crucial values or a statistical software package to calculate the p-value for the F-statistic.
- Decide whether to reject or fail to reject the null hypothesis by comparing the p-value to the alpha level, which is typically.05.
The null hypothesis is disproved and it may be said that there are significant differences between the group means if the p-value is less than.05. To ascertain which groups vary substantially in this situation, post-hoc tests like Tukey’s HSD or Bonferroni tests may be used.
One-way Although the requirements of normality and homogeneity of variance must be satisfied before the test is run, ANOVA is a potent tool for examining variations between groups in studies on language training. Results that are erroneous as a consequence of certain assumptions being violated.
Watch this video clip on YouTube about How to do a One-Way ANOVA in SPSS
PRACTICE IN SPSS
Input this sample data table into SPSS to run One-way ANOVA
| Participant | Direct Instruction | Indirect Instruction | Task-based Instruction |
| 1 | 78 | 65 | 82 |
| 2 | 87 | 72 | 79 |
| 3 | 80 | 71 | 84 |
| 4 | 75 | 68 | 81 |
| 5 | 83 | 73 | 86 |
| 6 | 79 | 66 | 83 |
| 7 | 86 | 70 | 80 |
| 8 | 72 | 67 | 81 |
| 9 | 81 | 74 | 85 |
| 10 | 84 | 69 | 82 |
Each participant is represented by a row in this table, and each instruction type is represented by a column. Each participant’s reading comprehension scores for each form of teaching are shown as values in the cells. For instance, Participant 1 achieved reading comprehension scores of 78 with direct instruction, 65 with indirect instruction, and 82 with task-based instruction.
PROBLEM-SOLVING
A language school is looking to compare the impact of two alternative teaching philosophies on learners’ capacity for listening comprehension. The second technique uses interactive exercises and group discussions, while the first method uses a conventional lecture-based approach.
40 intermediate-level pupils were selected by the school, and they were split into two groups at random. Group B got the interactive exercises and group discussions whereas Group A received the customary lecture-based approach. After 12 weeks of teaching, a listening exam was used to evaluate the listening comprehension abilities of both groups.
The school is interested in learning which kind of instruction helps pupils’ listening comprehension abilities the most. They also want to know whether the mean scores of the two groups vary significantly from one another.
Determine which teaching strategy is more successful by doing an adequate statistical analysis using the study’s data and interpreting the findings.
CHAPTER SUMMARY
The chapter on data analysis and hypothesis testing in the language instruction book, especially utilizing SPSS Statistics, is number eight. It discusses various data types, how to interpret descriptive statistics, and various hypothesis tests, such as the correlation and one-way ANOVA. Independent samples t-tests, paired samples t-tests, and one-sample t-tests are also covered. For language teachers looking to use data analysis and hypothesis testing to enhance their teaching methods, this chapter offers as a thorough reference. It explains how to carry out each test using SPSS Statistics and how to evaluate the results step-by-step. In order to demonstrate how these assessments may be used in language education research, the chapter also offers example data tables.
QUESTIONS FOR REVIEWING THE LESSON
- What are descriptive statistics and how are they used in studies on language teaching?
- Using SPSS Statistics, what kinds of data are there, and how are they analyzed?
- How are null and alternative hypotheses investigated using hypothesis testing, and what are the differences between them?
- What are the various t-test types, and how are they used in studies on language teaching?
- What is the one-way ANOVA and how does it compare studies on language teaching?
- What distinguishes ratio data from ordinal, nominal, and interval data?
- How is statistical significance determined? What is it?
- What distinguishes the paired samples t-test from the independent samples t-test?
- How does an SPSS correlation analysis work?
- What is simple linear regression and how is it used to studies on language teaching?
- What is a one-way ANOVA and how is it used in studies on language teaching?
- How do you examine data in study on language education using SPSS?
REFERENCES
Bitchener, J., & Knoch, U. (2009). The value of a focused approach to written corrective feedback. ELT journal, 63(3), 204-211. https://doi.org/10.1093/elt/ccn043
Dornyei, Z., & Ottó, I. (1998). Motivation in action: A process model of L2 motivation. Working Papers in Applied Linguistics, 4, 43-69. https://nottingham-repository.worktribe.com/output/1024190
Field, A. (2013). Discovering statistics using IBM SPSS statistics. Sage.
Gall, M. D., Gall, J. P., & Borg, W. R. (2007). Educational research: an introduction (8. utg.). AE Burvikovs, Red.) USA: Pearson.
Mackey, A., & Gass, S. M. (2021). Second language research: Methodology and design. Routledge.
Larsen-Freeman, D., & Anderson, M. (2013). Techniques and principles in language teaching 3rd edition-Oxford handbooks for language teachers. Oxford university press.
Larsen-Freeman, D., & Long, M. H. (2014). An introduction to second language acquisition research. Routledge.
McDonough, J., & McDonough, S. (2014). Research methods for English language teachers. Routledge.
Kottler, P., & Keller, K. L. (2009). Marketing management. Jakarta: Erlangga.
Pham, V. P. H. (2023). The Impacts of Collaborative Writing on Individual Writing Skills. Journal of Psycholinguist Research. https://doi.org/10.1007/s10936-023-09939-2
Salkind, N. J. (Ed.). (2010). Encyclopedia of research design (Vol. 1). sage.
Using SPSS Statistics specifically, Chapter 8 focuses on data analysis and hypothesis testing in the context of teaching languages. The chapter discusses several data types (nominal, ordinal, interval, and ratio), how to interpret descriptive statistics, and various hypothesis tests, such as the independent samples t-test, paired samples t-test, one-sample t-test, correlation, and one-way ANOVA. Overall,…