3. SKILLS LAB - Beginners guide: QNT T-Test
Katie Saunders
Created on July 25, 2024
Over 30 million people build interactive content in Genially.
Check out what others have designed:
Transcript
Micro-teach session 3: T-Tests
Beginners guide: Quantitative Research Methods.
Let's go!
Start course
Thank you for taking part in the STARS Quantitative research sessions. This session should take no longer than 10-15 minutes.This supplementary training is designed to further support you in gaining knowledge in quantitative research methods, and data collection. This module will explore:
- Paired samples T-Test
- Independant samples T-Test
Introduction
Micro-teach session 3: T-Test
NEXT
This training will only benefit those that understand basic quantitative research, this could be from previous STARS session or general knowledge, this module will not discuss basic principles such as types of data (nominal, ordinal etc.) or how to clean/ sort your data – this training begins after data is cleaned and ready to analyse. The data for this module has been created by the author (KS) and manipulated for the expected outcomes for each test, none of the data used in this module is real and therefore ethically sourced. All the scales used are also fabricated. The subject area for this module is based on psychosocial principles but this can be reflected across other clinical areas, the subject basis for this module is watching films and/or eating behaviours. You can access the data set for this module using this link: XXXXXXXXXXXXXX (This is a excel 365 document and just needs to be downloaded and imported into SPSS)
The purpose of this training is to provide a resource to support you in running a series of different statistical tests using SPSS [This module uses SPSS 29, but the process is similar across versions], this training is in line with Level 4 (first year university) introduction to research methods and statistics.
STARS SKILLS LAB
NEXT
Running a T-Test...
Once you have established your data is normally distributed this means that your are able to run a parametric test. (non-parametric tests will be discussed in a different micro-teach session).There are two types of T-Test:
- Paired samples T-Test: when you are looking a maximum of two variable within the same group.
- Independant samples T-Test: when you are looking at two variables that are separate from each other.
NEXT
PART ONE: PAIRED SAMPLES T-TEST
NEXT
Running a Paired Samples T-Test on SPSS
Example scenario:You are interested to see if watching movies can increase happiness. You have decided to ask a group of participants (n = 180) to help with figuring this out. These participants have filled in a happiness survey, they then watched a movie, and then they filled in another happiness survey. Now we will have a look how you run a paired-samples T-Test on SPSS, this is via the Analyse> Compare means and proportions > paired- samples
Hypothesis:You suspect that watching movies will increase happiness.
NEXT
Checking your outputs for significance....
Tip: the box you are looking at is the 'Paired Samples Test'.
NEXT
Reporting your findings.
The average score for happiness was higher after watching a movie than before watching a movie(64.94 compared to 25.65). A paired t-test showed that the differnences between the conditions was significant, (t = -17.03, df = 179, p = <.001, one-tailed).
One-tailed (one-sided p) = You already know the direction of the hypothesis. e.g. You suspect that happiness will improve after watching movies.Two-tailed (two-sided p) = You are not sure of the direction of the hypothesis. e.g. you are not sure if movies will increase happiness or not.
NEXT
Measures of effect size.
Now we have established that the findings are significant, we can now go and look at what the effect size is. Do do this was have to use the following calculation:d = (x1 - x2)/mean SD. It may look scary but it very simple to complete.
The mean for before movie was: 25.65The mean for after the movie was: 64.94
Step 1: Minus the one condition from the other (it is not important which mean you take from which)25.65 - 64.94 = 39.29Step 2: find the SD by adding the two SD figures together and dividing them by 2.8.15 + 33.97 / 2 = 21.06Step 3: Use the calculation....d = 39.29 / 21.06 = 1.86
Effect size according to Cohen (1988) is:
- Small effect size: 0.2 or more.
- Moderate effect size: 0.5 or more
- Large effect size: 0.8 or more.
Therefore we can update the write up to the following:The average score for happiness was higher after watching a movie than before watching a movie(64.94 compared to 25.65). A paired t-test showed that the differences between the conditions was significant, the size of the effect was large (t = -17.03, df = 179, p = <.001, one-tailed, d = 1.86).
NEXT
So, what did you want to find out? You wanted to know if watching movies could increase happiness. You asked a group of people to fill in a happiness survey before and after watching a movie. So, what did you do? You ran a paired samples t-test on SPSS, this test enables you to look at significant differences between two variables that are related to the same participant(s). In this case it was between the happiness scores for participants before and after watching a movie. So, what did you find out? The data revealed a significant difference in happiness scores before (25.65) and after (64.93) watching a movie (p <.001) [p = less than 0.050]. These findings suggest that matching movies can improve feelings of happiness.
So what?
NEXT
PART TWO: INDEPENDANT SAMPLES T-TEST
NEXT
Running a Independant Samples T-Test on SPSS
Example scenario: You now want to know what type of moive may influence happiness scores. You decide to ask people to report their happiness, watch either a a horror movie or a comedy movie. You then ask them to fill in another happiness questionaire.Now we will have a look how you run a independent-samples T-Test on SPSS, this is via the Analyse> Compare means and proportions > independent- samples Tip: When you are allocating your groups you need to remember what number you allocated to each group.
Hypothesis:You are not sure which movie will score higher.
**we will include both the before and after in the analysis but will only be looking at the after as we are looking at the between groups not the differences before and after watching a movie**
NEXT
Checking your outputs for significance....
Tip: the box you are looking at is the 'Independant Samples Test'.
Reporting your findings.
NEXT
The average score for happiness was higher for comedy after watching a movie than watching a horror movie.(73.32 compared to 57.10). A paired t-test showed that the differences between the conditions was significant, (t = -3.29, df = 178, p = .001, two-tailed).
One-tailed (one-sided p) = You already know the direction of the hypothesis. e.g. You suspect that horror will score higher than comedy.Two-tailed (two-sided p) = You are not sure of the direction of the hypothesis. e.g. you are not sure which movie will increase happiness more.
NEXT
Measures of effect size.
As with the paired-samples, we need to find the effect size:Using the same calculation:d = (x1 - x2)/mean SD.
The mean for a Horror movie was: 57.09The mean for a comedy movie was: 73.32
Step 1: Minus the one condition from the other (it is not important which mean you take from which)57.09 - 73.32 = 16.23Step 2: find the SD by adding the two SD figures together and dividing them by 2.33.37 + 32.75/ 2 = 33.06Step 3: Use the calculation....d = 16.23 / 33.06 = 0.49
Effect size according to Cohen (1988) is:
- Small effect size: 0.2 or more.
- Moderate effect size: 0.5 or more
- Large effect size: 0.8 or more.
Therefore we can update the write up to the following:The average score for happiness was higher after watching a movie than before watching a movie(64.94 compared to 25.65). A paired t-test showed that the differences between the conditions was significant, the size of the effect was small to moderate (t = -3.29, df = 178, p = <.001, two-tailed, d = 0.49).
NEXT
So what?
So, what did you want to find out?You wanted to know what type of movie may influence happiness scores. The chosen genres are horror movies and comedy movies. So, what did you do? You ran an independent samples t-test. This test enables you to look at differences between groups to see which group had a higher or lower scores. In this case it was to look at differences in happiness scores after watching either a horror movie or a comedy movie. So, what did you find out? The data revealed a significant difference in happiness scores between horror (57.09) and comedy (73.32) movies (p <.001) [p = less than 0.050]. These findings suggest that watching comedy genre movies are mean higher happiness scores than watching horror movies.
NEXT
Please provide your feedback for this micro-teach session...
Course completed!
You can now run two different T-Tests!You can now move onto micro-teach session 4...