June 23 - Inferential Statistics

We have now examined descriptive statistics. Today, we will take a close look at inferential statistics. Inferential statistics are used to determine if mean scores between two or more groups are statistically significant. There are several different types that we will examine that can be used with different numbers of groups for statistical comparison purposes.

The type of inferential statistic you will use depends on how many groups' mean scores you are comparing. The type also depends upon whether or not you will be using a pretest. In addition, if you are examining more than one type of variable, there is yet another type of inferential statistic for this.

Today, we will be examining the five different inferential statistics to compare mean scores. We will also learn about one type of inferential statistic that may be used to compare variances. Keep in mind that if variances are different (like we saw in our previous work in descriptive statistics), it is very difficult to compare the mean scores and make any sense of the comparison.

Whereas we calculate descriptive statistics in any quantitative design, we only use inferential statistics when we compare scores of two or more groups. Therefore, we would only use these if we are doing causal-comparative or experimental research since these are the only types of research where we make comparisons between groups. In essence, we do inferential statistics so that we may infer that what we did with one group worked better than what we did with the other(s).

We will be examining six different types of inferential statistics. These six include:

1. Levene's Test of Homogeneity of Variance - This is a type of inferential statistic that we use to compare variances (or spreads of scores in the different groups). We use it to make sure that there is no statistical difference between variances among groups before we compare their mean scores using one of the next inferential statistical procedures. It provides an "F' statistic.


2. Independent t-test - An independent t-test is used to compare the mean scores of two groups. It provides you with a "t" statistic.


3. ANOVA (analysis of variance) - An ANOVA is used to compare the mean scores of two or more groups. It provides you with an "F" statistic.


4. ANCOVA (analysis of covariance) - An ANCOVA is used to compare the mean scores of two or more groups, while considering the pretest results. It provides you with an "F" statistic.


5. Factorial ANOVA - A factorial ANOVA is used to compare the means scores of two or more groups when using two or more independent variables in your research (e.g., gender and reading method, grade level and math method, etc.). It provides you with an "F" statistic.


6. Bonferroni - A Bonferroni test is a specialized comparison of mean scores that may be used to isolate where statistically significant differences are occurring when you have found a difference in the mean scores of three or more groups (e.g., you have three reading methods and find a statistically significant difference among them. The Bonferroni adjustment allows you to identify where the specific differences are occurring. Is it between reading method one and two, or two and three, or one and three?). It is a specialized t-test. It provides a t-statistic.

Notice, on the above, that when you use an inferential statistic that allows you to compare mean scores of two or more groups, you will have an "F" statistic. When you use an inferential test that only allows for comparisons between two groups, it gives you a "t" statistic. How are they related? Like the relationship between standard deviation and variance, the relationship between "t" and "F" is the following: The square root of F = t, or said another way "t" squared is equal to "F."

One last word, keep in mind that once we calculate a statistic, we must ask the question, "Is this statistic statistically significant?" Just as we did with correlational research, we answer this question by looking at the p-value (going by other names such as probability, statistical sig., or simply sig.). Please return to Blackboard to begin this exploration of inferential statistics.

June 22 - Descriptive Statistics

We are now to the final part of research prospectus paper. This is the analysis part of the Design and Analysis subsection of the Method section of your paper. We have previously discussed one type of analysis, which is used in qualitative studies. This type is known as content analysis, which is where you create, revise, and label categories based upon narrative data. Today, we will be examining descriptive statistics, which form the basic statistics for all quantitative approaches.

Descriptive statistics include:

1. Measures of Central Tendency - these measures show how the scores in a set of scores "come together." Measures of central tendency include the mean, the median, and the mode.


2. Measures of Variability - these measures show how the scores in a set of scores "spread apart." Measures of variability include the range, the standard deviation, and the variance (the standard deviation squared).


3. Measures of Relative Position and Relationship - these measures include correlation coefficients, percentile ranks, and other such scores. They show relationships and relative positions of scores achieved.

These descriptive statistics form the basis for the next set of statistics that we will discuss - inferential statistics. Inferential statistics are used when statistical comparisons are made concerning, in particular, means and variances. Inferential statistics are only calculated in experimental and causal-comparative studies. However, descriptive statistics are necessary for any type of quantitative study whether it be descriptive, correlational, causal-comparative, or experimental.

Please return to Blackboard to begin this exploration of descriptive statistics. The information there will be of use to you whether you are doing descriptive, correlational, causal-comparative, or experimental research.

June 17 - Experimental Research

At this point, we have examined correlational and causal-comparative research along with the other designs of research. We have noted that neither causal-comparative nor correlational research designs can provide us with strong cause and effect information. We cannot say from the results of these types of studies that something caused something else with any sort of strong confidence. In fact, we would make a critical error in trying to do so.

Today, we will be examining our final type of research that does allow us to draw cause and effect conclusions. This type of research is experimental research. Experimental research is used extensively in the medical field and various other science-related fields. In addition, it is used in the testing of a variety of products in a host of fields. It is also used in education to try to examine cause and effect.

We can use experimental research when we ask such questions as:

• If we use this method of instruction would it work better than this other method as indicated by Benchmark scores?


• If we use two different methods of instruction, would one lead to higher levels of motivation?


• If we use this textbook with one group and use a different textbook with the other, will end-of-unit test scores be statistically significantly different?


• Would one management technique lead to fewer numbers of misbehavior in my class than another?


• Would students physical testing scores be different based upon which type of three possible types of training they used?

There are multiple other questions we might ask similar to these in design. These are just a few. In today's and Monday's examination of experimental research, you will be doing several activities to help you better understand this important design. Please return to Blackboard to begin.

June 16 Causal-comparative and Correlational Research

We have now examined qualitative and descriptive research. Keep in mind that these two types of research use methods that answer a research question that has been asked just after your review of related literature. Today, we will be looking at the types of research for which you will have a hypothesis that will be tested through the parts of your Method section.

We will begin this discussion with two types of research that are somewhat similar in some ways and different in others. We will be examining correlational and causal-comparative research. Keep in mind that a big difference in the two is that causal-comparative is designed to compare the scores of two groups based on some sort of fixed or non-manipulable variable (e.g., gender, intelligence, family background, SES, use of drugs, behavior disorder, etc.) whereas correlational is used to examine one group with two variables to see if these variables are related to each other.

When using either design, it is not possible to make cause and effect statements about the variables studied. For correlational, you only have one group so no comparison is occurring. For causal-comparative, many other variables may come into play when you try to study one certain one since you are studying these variables ex post facto (after the fact).

For instance, if you are comparing two groups of adults (one that uses drugs and the other that does not) on income earned, there may be other factors coming into play such as education obtained, family background, and depression. Any of these factors could potentially have impacted the result (Some people try to solve this problem by doing what is called "matching."). Therefore, conclusions from either correlational or causal-comparative are tentative. One cannot indicate cause and effect (correlational), and the other cannot control every variable that might be involved in a cause and effect since it is studied ex post facto. On the other hand, finding a relationship (correlational) or seeing a difference (causal-comparative) can be a first step toward doing an experimental research study.

Please proceed to Blackboard to begin examining these two types of research. You will find information there concerning these two types of research. Please proceed to the discussion board first before beginning the information on causal-comparative research. It will help you better understand the potential variables that may be examined through this type of research.

June 15 - Descriptive Research

Today, we will be examining descriptive research. Although different educational research experts use different designations for types of descriptive research, the different types can be loosely placed into a few specialized categories. These include:

• Survey Research (where participants are asked to complete a survey of some sort)


• Observational Research (where participants are observed and their behaviors are noted)


• Interview Research (where interviews are actually conducted)

In consideration of these three types, some people refer to interviews as "personal surveying" and place interviews under a "survey" heading. Others use the phrase "survey research" rather than descriptive research. Still others tend to shy away from using observation research. Yet others use the phrase "self-report studies" versus "observational studies." In any case, we can speak of these three different types realizing that various researchers consider them to overlap in some ways.

In any case, there are commonalities. For example, in descriptive research, whether you are doing survey, observational, or interview research; all data are reported quantitatively and are analyzed through typically fairly simple statistics. In addition, scoring techniques and various types of item categories are used to help organize the data into meaningful information. Some of the most common ways this is done is through:

• Likert Items (levels of agreement or disagreement with the statement made)


• Rating Items (items where you are typically asked to rate the usefulness or excellence of something)


• Semantic Differential scales (items that are split into bipolar ends [such as hot-cold, heavy-light, etc.] and where you make a mark toward one of these ends concerning whatever you are rating)


• Behavior checklists (used in observational research to see how often or how much a particular behavior occurs)


• Demographic items (to determine percentages of groups participating [e.g., census])

You have all participated at one time or another in a descriptive study. Perhaps you completed a restaurant survey. You participated in the US Census. You filled out a satisfaction survey. You used a shopping card from your favorite grocer. You did a peer observation, or you evaluated a professor. All of these are used for descriptive research purposes.

Please proceed to the Blackboard information to examine a scanned handout on some of the important aspects to consider when designing a descriptive research study. There you will find a brief history of survey research. You will also find information about different types of surveys and some of the problems that can exist with observational and survey research.

June 14 - Qualitative Research

At this point, we have been discussing what you all have in common with your research proposals. All of you have a title page. You will each have an Introduction section with an introductory portion, a statement of the problem, a review of related literature, and a hypothesis OR research question. In addition, you will all have a Method section with four subsections. We have discussed two of these. One subsection is your Participants portion where you let the reader/audience know who the participants in your research will be. The second subsection is your Instruments (for quantitative) or Data Collection Methods (for qualitative) portion. In this part, you will talk about what tools you will use to collect your data.

The final two subsections of your Method section are what we will be devoting the rest of our time to investigating. One of these is the Procedures subsection, and the other is the Design and Analysis subsection. These two subsections are highly related to one another. These two subsections are dependent upon the type of research design you choose (qualitative, historical, descriptive, correlational, causal-comparative, or experimental). So, in the next few days, we will be learning more about each of these different types of designs.

As noted on your syllabus, we will begin with the qualitative designs. Although there are a whole host of different qualitative designs, we are only going to focus on three of them. These include:

1. Case Study Design (This is the most common and is what you will be designing in this class if you choose to do a qualitative design.)


2. Phenomenological Design


3. Ethnographic Design

There is an activity in the Content portion of Blackboard in the Qualitative folder (June 14) where you are asked to look at chart with information concerning these three different qualitative designs. You are to examine overall similarities (i.e., what do they have in common?) and differences (How are they each unique?). Keep in mind that the tools for data collection that are mentioned could be used in any of the designs. Some just use them more commonplace than others.

It should become clear to you, as you consider qualitative designs, that they are quite a bit different and come from a different perspective than quantitative designs. We have already mentioned some of these. Another important way in which they differ is in how sampling occurs. In qualitative, the sampling procedures are purposeful instead of quantitatively controlled. You will see types of sampling used such as the following:

• snowball sampling (where you count on one trusted informant to lead you to others)


• deviant case sampling (where you are looking for people with a "deviant" behavior of some sort to sample)


• typical case sampling (where you are looking for people who generally express some behavior you want to study)


• opportunistic sampling (where you are looking for people anywhere you can find them as you go about examining some case, program, situation, etc.)

These are just a small sample of the types of sampling used in qualitative research. There are many others that go beyond the scope of this course. You can see that instead of trying to tightly choose and control a sample (into one group or another many times) like what is done in quantitative you, instead, are letting things occur naturally - including the sample.

As you can see, today's emphasis will be upon how a qualitative case study design works. Please proceed to Blackboard to begin examining qualitative designs to research.