THE ULTIMATE GUIDE FOR WRITING THE RESULTS SECTION IN YOUR DISSERTATION
THE RESULTS SECTION IN YOUR DISSERTATION
You have finally succeeded in completing the challenging task of conducting research for your dissertation. Depending on the path that you decide to take, this investigation can either be primary or secondary in nature. Bravo! You will now need to conduct an analysis of your data and write the section of your dissertation that deals with the results. If this describes you, then you've arrived at the proper spot.
When asked why it is so hard to complete a dissertation, the typical student will answer one or both of these questions. Either they despise analysing huge quantities of information or they do not enjoy producing lengthy amounts of text. It is laborious and uninteresting to do! Students shed tears.
Students, we hear you. This guide will help you create your results section for your dissertation. It is very detailed and helpful. So that you can focus on the most important things, we have separated the information into qualitative and quantitative results.
WRITE OUT YOUR QUANTITATIVE RESULTS
UNDERSTANDING THE FUNDAMENTALS OF RESEARCH
Before you can write accurate quantitative results, it is important to understand the basic principles of your research. It is a good idea first to recall the main variables or what you have analysed.
Every quantitative research project should have at least one dependent and one independent variable. At this stage, it is important to clearly define these variables. To assess the effect of an independent variable on a dependent variable, it is feasible to manipulate the result variable. Your outcome variable is therefore a dependent variable.
Next, decide whether your variables are categorical or continuous.
Categorical variables have a fixed number of possible values. Continuous variables can have a wide range of final scores. Remember to include any covariates or confounder variables. This variable may have influenced the connection between the dependent and independent variables. It would help if you controlled it to determine the relationship between your main variables accurately.
Choose a case study to demonstrate this. Imagine if your goal was to determine whether a relationship existed between height and self-esteem. Self-esteem is a dependent variable, while participants' height is an independent variable. It is possible that you also wanted to determine if there was a relationship between self-esteem and height after controlling for weight. You need to take into account weight in this instance.
Another illustration This might be used to find out if more men than women are interested in reading a certain romance book. Your gender serves as the independent variable, while your interest in reading the book serves as the dependent variable in this scenario. This variable is categorical since there are several forms of gender (male or female). On a scale from 1 to 10, indicate your desire to read the book. If you gave your participants a scale that ranged from one to 10, with one denoting that they had no interest to read the book and ten indicating that they had a great urge to do so, then this would be considered a continuous variable. However, this is a categorical variable if they were asked to identify whether or not they desired to read the books. "Yes" and "no" come in two different forms.
After controlling for the participants' relationship status, you might have wanted to see if there is a link between gender and willingness to read the book. The confusing factor is relationship status.
These illustrations will appear often throughout the blog post. It is critical to note that by organising your research in this way, you will be able to compose your results section fast and efficiently.
Let's move on to the following step.
1. A DESCRIPTION OF DESCRIPTIVE STATISTICS
Analyses that are intended to test your hypotheses should be reported using descriptive statistics and frequency statistics. These statistics may be used to summarise your data collection, either by concentrating on certain groups or by examining the full sample.
To be able to offer frequency and descriptive statistics, all variables utilised in your study must be listed, along with whether they are continuous or categorical.
Descriptive statistics may be used to characterise continuous variables. They provide both the variation or spread (standard deviation) and the central tendency (mean) measurements. You use frequency statistics to report the frequency (or number) of participants in each category, as well as percentages. These statistics both require that you create a table and must comment on the statistics.
What does this all look like in practice? These are just two examples. Your research will consist of three continuous variables if you have assessed the relationship between self-esteem and height and controlled for weight. The means and standard deviations of each variable must be reported in a table. Commenting on the outcomes is as easy as saying:
Participants averaged 173.50 cm in height (SD = 5.81) and 65.31 kilogrammes (SD = 4.44). Participants had average self-esteem (M = 5.55, SD = 2.67).
As you can see, if participants' self-esteem was graded on a scale from 1 to 10, this example indicates that their levels were moderate. The average level of self-esteem is moderate, with a value of 5, which is in the centre of the spectrum. If the median number was greater (for example, M = 8.33), you would infer that the participants' average self-esteem score was low.
This example shows how descriptive statistics should be reported for the entire sample. Descriptive statistics can also be provided for certain groups.
Let's go back to the second example. You want to report on the male and female desire to read romantic novels. This determination was based on a continuous scale of 1–10.
Let's now focus on frequency statistics. These are the numbers you will use to describe categorical variables.
From our second example, you can see how frequently data are provided for different groups. This is about gender, the desire to read romance books, and the relationship status of the individuals.
Three categorical variables would exist if participants were asked to choose "yes" or "no" as their response. You are reporting frequency and percentages rather than averages or standard deviations.
This is how you would describe it: You are keeping track of how many men and women desired to read the book, as well as how many were in relationships. The following formats are possible for reporting these statistics:
The book was requested by 20 (40%) male participants and 35 (70%) female participants. Also, 22 (44%) men and 26 (52%) women said they were in a relationship at the moment.
2. PROVIDING A REPORT BASED ON THE FINDINGS OF A CORRELATION ANALYSIS
After clearing it out, let's talk about how to present the findings of different statistical tests.
In order to determine if one or more continuous, independent variables are connected to one another, correlation is utilised as the first method (continuous, dependent). For example, you would be curious to see whether there is a relationship between the participant's height and their degree of self-esteem.
The very first thing that you need to do is report on whether or not your variables have a normal distribution. Examining a histogram that depicts your data will allow you to accomplish this goal. If the histogram displays a curve that looks like a bell, often known as a normal distribution (see the graph in purple below). If so, a Pearson correlation analysis should be used since your data are normal. You must include the p-value you calculated as well as the correlation coefficient, or r value, in this section. The correlation coefficient must be less than .05. This will demonstrate significance. In the case that a correlation is found, you must mention anything along these lines:
The results of the Pearson correlation analysis indicated that there was a positive relationship between people's height and their self-esteem (r =.44, p .001); this was the conclusion reached by the study.
Positive correlation may also happen when higher levels of one variable are connected with higher levels of higher variables with high degrees of correlation. However, a negative correlation occurs when a variable's greater values are connected with its lower levels.
Not least, it's critical to keep in mind that p values higher than.000 should never be reported. Instead, use p =.000 and p =.001 to report it. Due to the fact that the p-value cannot exactly equal 0.000, this is crucial. The p-value will be shown in all other circumstances as "p =.011".
Performing a Spearman correlation analysis is indicated if your data are skewed or not regularly distributed (see red graphs). To report the results, you may write here.
Height and self-esteem have a positive relationship, according to a Spearman correlation analysis. (r, =.44, p >.001). You have used partial correlation if you used a covariate like a participant weight. After controlling for weight, your results will show you how closely participants' self-esteem and height correlate. After controlling for weight, the results are =.39, p =.034).
Additionally, you must prepare a table summarising your key findings. If you do not utilise covariates, you will obtain a straightforward table. Your table will be significantly more sophisticated if you utilised a covariate. Correlations that have previously been recorded in the table are denoted by the notation "-" in both tables. Additionally, "*" and "**" may be used to denote correlations at various degrees.
3. REPORTING ON THE RESULTS OF A REGRESSION
Reporting on regression analysis might be more challenging since you need to state if all assumptions have been satisfied (especially if you are writing a graduate dissertation).
To make sure that all presumptions are true, you should address these issues.
(1) You must demonstrate that your tolerance statistics are not less than.01 and your VIF statistics are not more than 10 in order to presume that there is no multicollinearity (i.e., that your independent variables are not highly correlated).
(2) to presumptively rule out any residual autocorrelation (i.e., no correlation between two observations). Depending on the number of participants (independent variables) and the number of predictors, your Durbin-Watson statistics are within the desired range (independent variables).
(3) You must analyse the scatterplot using standardised predicted values and come to the conclusion that your graph does not funnel out or curve in order to satisfy the assumptions of linearity (i.e., that independent and dependent variables have a linear relationship) and heteroscedasticity (i.e., because the independent variable's independent variable should have equal error term variance).
All of this could seem challenging. Actually, it's fairly easy. Simply take note of the Durbin-Watson and Tolerance values, the Durbin and Wilson values, and the scatterplot when you look at the results output. When your presumptions are correct, you may write:
Since none of the VIF values were lower than 0.1 or greater than 10, the presumption of no multicollinearity was satisfied. The range of Durbin-Watson statistics was as anticipated. This suggests that the presumption of no residual autocorrelation was also fulfilled. The linearity and homoscedasticity presumptions were satisfied since the scatterplot of the standardised residuals on the standardised expected values did not curve or funnel out.
You need to look deeper into the meaning of your assumptions if they are not being met. Andy Field wrote a chapter about regression, especially the section on assumptions. It is a good idea to read it. Download his book now. Everything you need to know about regression analysis assumptions, including how to test them and what to do if you don't satisfy them, is covered in this book.
Let's now concentrate on the actual findings of the regression analysis. Let's say you want to find out whether, after adjusting for weight, participants' height predicts their level of self-esteem. You've expanded the model by including weight and height as predictors. Self-worth is a relying factor.
The model's significance in forecasting self-esteem scores must first be reported. Examine your ANOVA analysis's results, and record the residuals' degrees of freedom, the model's F value, and both. the degree of significance, as well.
The percentage will be 33.5 percent if your R2 score is 33.
Report the coefficients of your model. The Coefficients table will be included in your output. Also, note a β value. t Each predictor has a value and a significance level.
The significance of your predictor is measured by its value. This could be, for example, whether participants' self-esteem scores were predicted by their height. Comment on the value of β. This result corresponds to a unit increase of the predictor. As a result, if the participants' height is the predictor/independent variable and your b value is.351, it follows that self-esteem improves by.35 for every 1 cm of height increase. The weight of the contestants is also accurate.
Your model also included weight and height as predictors. Participants' self-esteem was affected by height, even after weight control.
4. REPORTING RESULTS FROM A CHI-SQUARE ANALYSIS
We have shown that only continuous variables allow for regression and correlation. You'll see that when there are continuous and categorical variables, t-tests, ANOVA, and MANOVA may be used. What we'll discuss is chi-square analysis. When all of the variables are categorical, it is applied.
To ascertain if gender—a categorical dependent variable with two levels: men and females—influences the choice to read a book, chi-square analysis can be performed. By responding to yes/no questions (categorical dependent variables with two levels, yes/no), this variable is evaluated.
The findings of a chi-square test must be reported. This entails taking note of the degrees of freedom, significance values, and Pearson Pearson chisquare value. You may see all of them in your output.
5. REPORTING RESULTS FROM A T-TEST ANALYSIS
Let's now talk about how to present the results of a t-test. This test examines if significant differences exist between two participant groups. Your dependent variable is continuous, but your independent variable is categorical (for instance, gender) (e.g. determination to read the book as assessed on a 1-to-10 scale). In this case, you're trying to figure out if men or women are more determined to read love books.
You may remember that you had previously given these variables descriptive statistics. You took note of the average scores for men and women on the decision to read the book in this case, as well as their standard deviations. The t value, degree of freedom, significance level, and other details must now be reported. Your output will show all of this.
You could say:
Males had a lower ability to read the novel than did females (M = 3.20; SD =.43). A t-test analysis showed that this difference was significant (t(54), = 4.47, p.001).
6. REPORTING ONE-WAY ANOVA RESULTS
One-way ANOVA is used when there are more than two comparison alternatives or when there are several occurrences of a continuous independent variable and a categorical dependent variable.
A categorical dependent variable has two conditions in the t-test example. Depending on whether the participant was a woman or a man, this was the case. To determine someone's relationship status (an independent variable that includes three levels of single, married, and divorced), it would be necessary to consider three factors.
The findings would be presented similarly to a t-test. The means and standard deviations for each participant group should be reported first. Then you list the people with the highest and lowest means. The outcomes of the ANOVA test will then be shown. This includes the F value, the degrees of freedom (both within and between individuals), and the significance value.
It's crucial to bear in mind two things. It is crucial to review your output and establish if the Levene's Test is significant before you disclose your findings. This test quantifies the variance or homogeneity. The presumption is that the variance should be the same across all comparison groups. You must provide the usual F value if the test findings are not statistically significant.
If the test is significant, you must supply the Welch statistic, associated degrees of freedom, and significance level.
Only if there were noticeable differences between groups may be determined using an ANOVA. It does not, however, indicate where the discrepancies are. A post-hoc analysis (Tukey's HDT test) will be required. You can see which comparisons were meaningful from this output.
The way you should submit your findings is as follows:
The people who were unmarried (M = 7.11, SD =.45) and those who had recently divorced (M = 5.11, SD =.55) were the most committed to reading, respectively. (M = 4.95, SD =.44). Significant group differences were found using ANOVA (F (2,12) = 5.12, p =.004). Post-hoc comparisons between individuals in relationships (=.003) and between single participants and those who had divorced (=.004) showed significant differences. Participants in relationships and divorcees did not vary significantly from one another (p =.067).
7. PROVIDING A REPORT ON THE RESULTS OF THE ANCOVA
If you wish to evaluate the main and interaction effects of categorical factors on a continuous dependent variable while simultaneously accounting for the impacts from other continuous variables, you should use the statistical approach known as analysis of covariance (also known as ANCOVA) (or covariates).
After adjusting for participants’ general interest in reading, you may use ANCOVA for testing, for example, whether relationship status (a continuous dependent variable, rated on a 1-10 scale) influences the decision to read a romance novel. After adjusting for the participants' general interest, you will use ANCOVA to test if their relationship status affects the decision to read romance novels (continuous covariate). This is a scale of 1-10.
When you are prepared to present the findings, have a look at the table in your output that is titled "test of between-subjects’ effects." Report the major independent variable, covariate's F values and degrees of freedom for each variable and error. A significant ANCOVA, like an ANOVA, does not reveal the source of the differences in data. You must do deliberate comparisons in order to achieve this, and you must then provide the important values associated with the numerous contrasts.
You are able to report the findings in one of the following ways:
In this study, there was a significant relationship between the desire to read a romance novel and the covariate of general book interest (F(1,26) = 4.96, p.001). After adjusting for the effects of the participant's general interest in literature, there was still a significant relationship status effect on the choice to read the romantic book (F(2,26) = 4.14, p.001). The comparisons showed that in contrast to being in a relationship (t(26) = 2.77, p =.004) and divorced (t(26) = 1.89, p =.003), being single significantly increased the willingness to read the book. This was the conclusion drawn from the data analysis.
8. PROVIDING A REPORT ON THE OUTCOMES OF THE MANOVA
The multiple-group analysis of variance, or MANOVA, is the last test that will be covered in this tutorial. This test is used to determine if there are any changes in the continuous dependent variables between independent groups.
MANOVA (dependent variable) can be used to determine whether male or female participants (the independent variables) are more interested in reading romance books (the dependent variable), and crime novels (the independent variables).
Before presenting the results, you must first determine whether or not the so-called Levene's test and Box's test are significant. These tests evaluate two hypotheses: first, that the covariance matrices are equal (Box's test), and second, that the variances for each dependent variable are the same (Levene's test).
In order to determine whether or not your assumptions are correct, it is necessary for both tests to provide insignificant results. If the results of the tests are substantial, you will need to do more research to comprehend what this indicates. To reiterate, you may find it useful to read the chapter on MANOVA that was written by Andy Field and can be found here for your reference.
After this, you will need to submit your descriptive statistics in the manner that was shown to you before. You are going to report the means as well as the standard deviations for each dependent variable, and you are going to do so individually for each set of individuals. The next step is for you to look at the findings of something called "multivariate analysis."
You will see those four statistical values together with their related F and significance values provided. They have been designated as Pillai's Trace, Wilks' Lambda, Hotelling's Trace, and Roy's Largest Root, in that order. Using these statistics, you may determine whether or not your independent variable has an effect on the variables being assessed. Since this is standard procedure, just the Pillai's Trace is often reported. When presenting the results, you should include the F value, the degrees of freedom (both for the null hypothesis and the error), and the significance level, exactly as you would when publishing the results of an ANOVA analysis.
The value of one of the four statistics that were shown previously in this paragraph must also be reported, however. The statistic known as Wilks' Lambda, the statistic known as Hotelling's Trace, the statistic known as Roy's Largest Root, and the statistic known as Pillai's Trace may all be designated with the letter V, the letter A, or the symbol (but you need report only one of them).
The results of the tests that were conducted to compare the participants must also be considered (which you will see in your output). The research results of these tests will reveal the independent variable's individual effects on each of the dependent variables. The same format that you used for the ANOVA will be used to present these results.
WRITE UP YOUR QUALITATIVE RESULTS
HAVING A FUNDAMENTAL UNDERSTANDING OF YOUR RESEARCH IS ESSENTIAL
Because you won't have to deal with statistics, reporting qualitative findings is substantially simpler than publishing quantitative results. Consequently, this part is much condensed.
Prior to reporting the results of your qualitative research, you must reflect on the other types of research that you have previously conducted. Interviews, observations, and focus groups are the three most common forms of qualitative research, and it is likely that your study will fall into one of these categories.
The findings of all three categories of research are reported in essentially the same way. In spite of this, it might be beneficial for us to concentrate on each of them in turn.
1. REPORTING ON THE RESULTS OF INTERVIEWS
The qualitative data was analysed using theme analysis if semi-structured interviews were employed. As a result, it was up to you to read through the transcripts of the interviews, code individual phrases, and put codes in groups to create themes.
Consider, for example, if the subject of your qualitative study was what motivates young people to smoke. You asked your participants why they started smoking, why they kept smoking, and why they wanted to stop. Due to the way your study was designed, you already have three major themes: (1) the causes of starting to smoke, (2) the causes of continuing to smoke, and (3) the causes of quitting. After that, you examine the factors that led your participants to start smoking, keep smoking, and decide to stop. Every chosen cause will serve as a subtheme.
When reporting the findings, the language should be structured into subsections. Each section should handle a specific topic. Then, inside each part, you must discuss the subthemes that your data revealed.
Let's say you found that young people started smoking because they believed it was cool, were under peer pressure, observed their parents smoking, were less stressed when they smoked, and wanted something different. Your "reasons why people started smoking" theme now has five subthemes. You must now offer your results for each subtheme along with quotes that best capture that subtheme. This is carried out for every theme and subtheme.
Developing a table with all of your themes, subthemes, and related quotes is also a smart idea.
Here is an example showing how to cite a quote in text:
Many participants claimed that they began smoking because they believed it to be fashionable. One participant said, "I was just 15 years old at the time." Many participants claimed that they began smoking because they believed it to be fashionable. One participant said, "I was just 15 years old at the time." I was focusing on some older men that everyone thought were cool. I wanted more attention because I was shy. I reasoned that if I started smoking, I would resemble these older men. (Interview 1, Male)
2. REPORTING ON THE OBSERVATIONS
If your research involved observation, your objective would have been to watch one behaviour in a particular environment. Imagine you were observing a therapist telling patients that their symptoms were not physical but rather psychological. Before reporting the observations, you must first classify them.
You may have seen, for instance, that the therapist is eager to discuss the problem's origin, the patient's lack of medical issues, the patient's experience with stress, the link between stress and the problem, and the new knowledge. You can use these as the foundation for your observations.
Each subject requires a separate report. To accomplish this, first, describe your observation (such as a conversation or behaviour) before making a comment about it.
Here's an example:
Therapist: What was it that made you feel stressed in the past few months?
Patient: Of course. It was my belief that I would lose my job, but it didn't happen. My girlfriend and I split up shortly after that. But I was fine with these two things.
Therapist: Did your symptoms change when you were stressed versus when you were not?
Patient: Hmmm. Yes. They were present most of the time when I was going through those periods, now that I think about it.
Therapist: Could it be stress that is intensifying your symptoms?
Patient: It's something I had never considered. It seems quite logical to me. It is?
The therapist tried to establish a link between stress and the patient's symptoms in this case. The therapist did not tell the patient that stress was causing his symptoms. Instead, she guided him through questions to find the connection between his symptoms and stress. The patient is now able to see the connection between stress and his symptoms.
3. REPORTING ON THE RESULTS OF THE FOCUS GROUP
Focus groups are done similarly to interviews, with the exception that more individuals participate in them than one person does at a time. Focus group data analysis is quite similar to interview data analysis. Referring back to the earlier section on reporting interview results could be helpful.
Let's take, for example, the focus group that you conducted to identify why people prefer Schweppes to Coca-Cola. You have taken notes from your focus groups and extracted the themes. There are many reasons people choose one of these drinks over the other.
Your results report should include two sections. The first part outlines your preferences for Coca-Cola, while the second outlines your preferences for Schweppes. Each preference in each area has to be supported by an explanation. These exact justifications must to be related to specific quotations.
Here's an example:
Participants preferred Schweppes to Coca-Cola because Schweppes is less sweet. This idea was shared by several participants. One participant stated that Coca-Cola is too sweet and that he doesn't like it. "Schweppes has a more bitter taste, and I don’t feel like my stomach is being stuffed with sugar." (Participant 2, female). Another participant agreed, noting that she agreed with her comments. Coca-Cola tastes sweet and makes me feel sick. It is refreshing to drink a glass of Schweppes. "I don't feel like I need water after it" (participant 4, male),
CONCLUSION
You will feel like you have completed a marathon when you finish the results section of your dissertation. It feels amazing, does not it? Writing qualitative results is much easier than writing them. However, reporting statistics isn't difficult if you have a guide to assist you. This guide should have eased your concerns and given you the confidence to write your results section without too much difficulty.
