USA: +1 508 290 1604
USA: +1 508 290 1604

Responses to the assassination of Rev. Dr. Martin Luther King

Topic: Responses to the assassination of Rev. Dr. Martin Luther King

Data: NORC April 1968

Objectives:

  • Calculate and interpret descriptive statistics
  • Apply your knowledge of levels of measurement to public opinion survey variables.
  • Become familiar with using a survey codebook

Note on statistics commands:

After recoding some variables, you will need to run a frequency table to see the distribution of a categorical variable, calculate descriptive statistics for individual variables. This assignment can be completed in Excel or any statistics program. Instructions for R, STATA, and SPSS/PSPP are provided below.

Worksheet Instructions:

First, open up the data file in the statistics program you are using (or Excel).

Second, open the codebook, which is a Word document and find the question and answers for

Q.54. IN GENERAL, WOULD YOU SAY YOU HAVE BECOME MORE FAVORABLE TOWARD RACIAL INTEGRATION IN RECENT YEARS, OR ARE YOU LESS FAVORABLE TO INTEGRATION

  1. What is the mode for Q54?
    1. 1 – Became more favorable
    2. 2 – Became less favorable
    3. 3 – About the same
    4. 4 – No answer

The value categories for nominal variables are unordered. For ordinal variables, the categories are ordered, such as from smallest to largest in a clear progression.  In this format, Q54 is nominal because the categories are not in a logical or inherent order. The best way to transform Q54 to make it an ordinal variable would be to tell the computer to exclude (or declare missing) “4 – No answer,” and switch the order of “2- Become less favorable” and “3 – About the same.” To do this, follow the instructions to recode the variable, and run a frequency table or descriptive statistics to identify the median value.

  1. What would be the median value of a reordered Q54?
    1. 1 – Became more favorable
    2. 2 – About the same (recoded)
    3. 3 – Became less favorable (recoded)

Now look at  Q57, a question that asks for the survey respondent’s response to the assassination of Rev. Martin Luther King (d. April 4, 1968).

  1. What level of measurement is Q57?
    1. Nominal
    2. Ordinal
    3. Interval/Ratio
  2. What is the mode of Q57?
  3. 1 – Sadness
  4. 2 – Anger
  5. 3 – Shame
  6. 4 – Fear
  7. 5 – Felt he brought it on himself.

Exclude all respondents who responded with 6 – Don’t know or didn’t answer the question (7). In STATA, PSPP/SPSS, and R, this requires you to declare these values to be missing or complete a recode command.

In R-Studio, use car::recode to create a new variable called Q57R from Q57:

norcsrs4050_short$Q57R<-car::recode(norcsrs4050_short$Q57,”1=1;2=2;3=3;4=4;5=5;else=NA”)

In STATA, the best way to do this is to create a new variable with the name Q57R (R for recoded, but you can choose any new variable name you would like) with the command gen or generate:

gen Q57R = Q57 if Q57<6

This command tells the computer to create a new variable equal to the old variable as long as the old variable is less than 6. You would then use the new variable, Q57R, in all future analyses.

In PSPP the quickest way to exclude a value category is to declare the value as “missing.” To do this, enter the following command in the syntax window before any FREQUENCY or DESCRIPTIVE command:

missing values Q57 (6).

Always remember to include the period at the end of every PSPP command.

  1. What percentage of respondents (after excluding those who didn’t know or didn’t answer the question) indicated “1 – Sadness” was their response to MLK’s assassination? ______
  2. What percentage of respondents (after excluding those who didn’t know or didn’t answer the question) indicated “5- Felt he brought it on himself” was their response to MLK’s assassination?

Next, look at response to QS1A, QS1B, QS1C, QS1D, QS1E, QS1F. In a spreadsheet view, these questions should appear right after Q57. Each of these questions are measured the same way, in response to a question that said:

NOW I’M GOING TO READ SOME WAYS THAT PEOPLE HAVE FELT SINCE THEY HEARD THAT DR. MARTIN LUTHER KING, JR. WAS DEAD, AND I’D LIKE YOU TO TELL ME WHICH OF THE WAYS ON THIS CARD COMES CLOSEST TO YOUR OWN FEELINGS. READ EACH STATEMENT AND CIRCLE ONE CODE ON EACH LINE.

The card included the following four options for respondents to choose from:

  1. FELT THIS VERY STRONGLY
  2. FELT THIS FAIRLY STRONGLY
  3. CROSSED MY MIND
  4. NEVER OCCURRED TO ME

You will need to declare as missing (change to blank or a period) all responses of 5, “no answer.”

The first response queried was:

Q.S1A. FELT SORRY FOR HIS WIFE AND CHILDREN.

Once you exclude ‘5’ from the analysis, this variable is ordinal since the strongest emotional response is the lowest value category and the responses indicate diminished emotional strength as the categories ascend.

You will need to declare as missing (change to blank or a period) all responses of 5, “no answer.”

  1. What is the mode for QS1A? In other words, did the largest number of respondents feel sorry for his wife and children very strongly, or fairly strongly? Or, did the largest number of respondents say it merely “crossed their minds” or “never occurred” to them?
    1. 1- Felt this very strongly
    2. 2- Felt this fairly strongly
    3. 3- Crossed my mind.
    4. 4- Never occurred to me
  2. Since QS1A is ordinal, we can calculate and report the median, the mean and other descriptive statistics. What was the mean response? ____________
  3. Is it possible that given the size of the sample, that the mean value for QS1A would be quite different than the actual mean value of the population? The standard error tells researchers how accurate the mean of a sample is likely to be compared to the true population mean. In other words, how accurate is this calculation is the mean compared to what may be the actual mean value in the entire population of the USA in 1968. What is the standard error of the mean of QS1A? (see Quirk 2000, pages 3, 9-11 for instructions to calculate this in Excel, or see https://www.students4bestevidence.net/blog/2018/09/26/a-beginners-guide-to-standard-deviation-and-standard-error/ )
    1. 2
    2. 3
    3. 78
    4. 1
  4. What is the standard deviation of QS1A?
    1. 3
    2. 78
    3. 04
    4. 2
    5. 2
  5. What is the interquartile range of QS1A?
    1. 1
    2. 2
    3. 3
    4. 4
  6. What do these measures of dispersion tell us about the distribution of opinions?
    1. The measures of dispersion are somewhat small, indicating that most responses are similar to the mean and/or median values of QS1A.
    2. The measures of dispersion are rather large, indicating that the most responses are disimilar to the mean and/or median values of QS1A.
  7. True or False: At least half of all respondents felt sorry for MLK’s wife and children very strongly or fairly strongly.

Calculate the mean, median, mode, standard deviation and skewness for QS1B, QS1C, QS1D, QS1E, QS1F.

  1. Which of the responses to MLK’s death was felt most strongly, on average, by respondents? In other words, which of the following variables has the lowest mean value?
    1. QS1B Felt angry at the murder
    2. QS1C Thought about how this is like the many tragic things that happened to African-Americans
    3. QS1D Worried about how… this would affect your own life, job and future
    4. QS1E Felt angry at the rioters
    5. QS1F Worried at how much more violence there would be this summer
  2. Low levels of dispersion, like standard deviation, indicates that most responses were close to the mean or median value. In other words, this is a measure of how much people agree with the mean or median response. Which of the following variables has the lowest standard deviation, indicating that most response values were close to the mean?
    1. QS1B Felt angry at the murder
    2. QS1C Thought about how this is like the many tragic things that happened to African-Americans
    3. QS1D Worried about how… this would affect your own life, job and future
    4. QS1E Felt angry at the rioters
    5. QS1F Worried at how much more violence there would be this summer
  3. QS1F “Worried at how much more violence there would be this summer” should have the lowest standard deviation of the variables measuring people’s responses to the assassination other than QS1A. What was the median value for QS1F?
    1. 1- Felt this very strongly
    2. 2- Felt this fairly strongly
    3. 3- Crossed my mind.
    4. 4- Never occurred to me
  4. Which of the following statements best characterizes the pattern of responses for QS1F compared to other variables in this battery of responses?
    1. America in 1968 was quite united in both how it mourned for King and in how Americans viewed the political implications of his death, especially in their responses to the rioters who rioted after King’s death
    2. With the exception of a small group of people, most Americans shared a strong sense of grief after King’s murder and initially there was even a broad consensus towards political implications and in how people viewed people who rioted after King’s death..
    3. In a country divided in its responses to King’s assassination, there was greater consensus among Americans over shared concerns about violent implications of King’s death than there was a shared sense of grief over his murder.
    4. In a country divided in its responses to King’s assassination, there was more agreement in America about shared grief and especially over concerns for King’s family, than there was agreement over the implications of his death, especially towards those who rioted in anger after King’s death.
  5. Which variable has the highest IQR, indicating the least amount of agreement or consensus in responses to the question?
    1. QS1A Felt sorry for his wife and children
    2. QS1B Felt angry at the murder
    3. QS1C Thought about how this is like the many tragic things that happened to African-Americans
    4. QS1D Worried about how… this would affect your own life, job and future
    5. QS1E Felt angry at the rioters
    6. QS1F Worried at how much more violence there would be this summer
  6. Which variable has the lowest IQR, indicating the highest amount of agreement or consensus in responses to the question?
    1. QS1A Felt sorry for his wife and children
    2. QS1B Felt angry at the murder
    3. QS1C Thought about how this is like the many tragic things that happened to African-Americans
    4. QS1D Worried about how… this would affect your own life, job and future
    5. QS1E Felt angry at the rioters
    6. QS1F Worried at how much more violence there would be this summer
  7. More respondents answered “1” felt this strongly in response to QSE1 “felt angry at the rioters.” No other response was so popular to any of these questions. QSE1 also has the highest positive skew of these battery of variables. What sentence best explains why that variable has the highest positive skew?
    1. The variable with the highest mode has the highest positive skewness score because there are many observations at the two extreme ends of the scale and few observations in the middle categories.
    2. The variable with the highest mode has the highest positive skewness score because the there are so few observations in the other categories.
    3. The variable with the highest mode has the highest positive skewness score because the mode is value category 1, and as a result, most of the observations are far from the mean value which reflects a the presence of quite a few responses at the opposite extreme (4).
    4. The variable with the highest mode has the highest positive skewness score because the median is much lower than the mean value.

Instructions for Statistics Programs

R / R-Studio

In R, you will need the psych, RCPA3 and car packages. Install those packages and then you will need the following commands to make those packages active:

library(psych)
library(RCPA3)
library(car)

To see a frequency table for a variable (example is of Q24 using the dataset norcsrs4050_short), use RCPA3’s freqC command. Change the plot to = TRUE if you want to see a bar chart

freqC(norcsrs4050_short$Q24, digits=2, plot = FALSE)

To calculate the mean, standard error of the mean, standard deviation, median, interquartile range (iqr) and skewness of a variable, use psych’s describe command (psych::describe).

describe(norcsrs4050_short$Q24, IQR = TRUE) For everything except standard error of the mean, you can use RCPA3’s describeC command.describeC(norcsrs4050_short$Q24)

Before you calculate these descriptive statistics, you may need to declare certain values to be missing, to exclude them from the analysis and any calculation. Car’s recode package is the most straightforward (see lengthy description and examples on Quercus). To tell STATA that for variable Q24, the value 9 should be replaced with a missing value in a new variable called Q24r, use the command:

norcsrs4050_short$Q24r<-car::recode(norcsrs4050_short$Q24,”1=1;2=2;3=3;4=4;5=5;6=6;9=NA”)

To change a value, the list of values above uses the format old value=new value with each value separated by a semi-colon. So, if you need to change a value, change the number on the right side of the equal sign.

STATA

Completing the assignment in STATA simply requires the following three commands:

To calculate the mean, standard error of the mean (‘semean’), standard deviation (sd), median, interquartile range (iqr) and skewness (sk) of a variable, you can use the command ‘tabstat’. Here is an example using variable Q24:

tabstat Q24, s(mean semean sd median iqr sk)

To use this command to find these descriptive statistics for the variables you need for this assignment, simply replace “Q24” with the variable (or variables in a list) you want to use.

To see the frequency table for Q24 use the following command:

tabulate Q24

For short, you can use tab Q24 or tab1 Q24

The latter is useful when you want frequency tables for multiple variables with one command, like tab1 Q21 Q24 Q17-Q20 (the same works for tabstat)

Before you calculate these descriptive statistics, you may need to declare certain values to be missing, to exclude them from the analysis and any calculation. To tell STATA that for variable Q24, the value 9 should be replaced with a missing value, use the command:

replace Q24=. if Q24==9

For an entirely new variable that excludes values, use generate (or gen for short) like:

gen Q24R=Q24 if Q24<9

To change a value, you can use the replace command. To change Q24R value=2 to a 3, do the following:

replace Q24R=2 if Q24R==3

You can also use STATA’s recode command. This example leaves observations of variable Q24 with the value of 1 unchanged, changes 2 to 3 and values 4 through 6 to 4, and 9 into missing, using the format ‘old value’=’new value’, and creates new variable, Q24R rather than rewriting over the original variable.

recode Q24 (1=1)(2=3)(3=2)(4/6 = 4)(9=.), gen(Q24R)

PSPP

Go to the Analyze menu and click “Descriptive Statistics->Descriptives…” Select a variable on the left and click the button so that the variable name appears in the top-right corner. In the bottom-right corner, de-select minimum and maximum, select “standard error of mean,” and “skewness.” The click “Paste” and a new window appears called the Syntax Editor with the following text (in place of Q24, you should find whatever variable you selected):

DESCRIPTIVES

/VARIABLES= Q24

/STATISTICS=MEAN STDDEV SEMEAN SKEWNESS.

You can manually add variable names to the second line OR copy-and-paste the command to repeat the command.

Notice that median is not an option. For that, go to the Analyze menu and click “Descriptive Statistics->Frequencies…” This command will give you an entire frequency table. Choose a variable from the list on the left and click the arrow in the middle so that the variable name appears in the top-right. In the bottom-right corner, de-select minimum and maximum, select “standard error of mean,”“skewness,” and “median.” Then click “Paste” and the following will appear in the syntax editor window after the Descriptives command (in place of Q24, you should find whatever variable you selected):

FREQUENCIES

/VARIABLES= Q24

/FORMAT=AVALUE TABLE

/STATISTICS=MEAN STDDEV SEMEAN SKEWNESS MEDIAN.

Just like ‘Descriptives,’ you can manually add variable names to the second line OR copy-and-paste the command to repeat the command. When you are ready, go to the “Run” menu of the syntax editor window and click: “All”.  You will see your results in the output window.

If you need to declare missing values, you should put the missing values command before commands like Descriptives and Frequencies in the syntax window.

missing values Q24 (9).

Note: PSPP knows to run a command only when it sees a period. You can put a command on one line, or many lines as long as there are no periods. So, the periods at the end of the above commands are crucial

To change values, use the RECODE command. This example leaves observations of variable Q24 with the value of 1 unchanged, changes 2 to 3 and values 4 through 6 to 4, and 9 into missing, using the format ‘old value’=’new value’, and creates new variable, Q24R rather than rewriting over the original variable.

RECODE Q24 (1=1)(2=3)(3=2)(4 THRU 6 = 4)(9=SYSMIS) into Q24R.

Still stuck on your due assignments?
Hire our experts now and get it delivered within hours!
Order now