Hi, I am trying to find a way to find out the number of people in Maryland who earn $15 or less. I am using the analyze online tool. I have found other forums on the topic, but it has not helped. I am using the row variable “hourwage()”, the column is “year(2023)”; the filters are “statefip(24)”, “hourwage(1-15)”, and “pernum(1)”. It seems I’m getting a number that’s too large. Any suggestions would be greatly appreciated.
Thanks
You can use our online data analysis tool with IPUMS CPS data to estimate the number of workers in Maryland whose hourly wage is less than or equal to $15. See our brief video tutorial to learn how to use the online data analysis system.
The row and column fields are where you specify the variable(s) you want to tabulate. The filter field is where you make restrictions on the observations (individuals or households) you want to include in the tabulation.
Your estimate may be larger than expected for a couple reasons. First, you may be misspecifying your inputs. Second, you may need to adjust your approach to account for pooling multiple samples. The CPS is a monthly panel survey. Every year, there are 12 Basic Monthly Survey (BMS) samples and one Annual Social and Economic Supplement (ASEC) sample. If you are pooling multiple samples together, you need to adjust the sampling weight variable by dividing it by the number of pooled samples. For example, if you are using all the BMS samples from 2023, that is 12 samples. To obtain an annual estimate from those 12 BMS samples, you would need to divide the sampling weight variable (in your analysis, EARNWT) by 12. If you do not adjust the sampling weight to account for pooled samples, your estimates of frequencies will be much larger than expected.
Our online data analysis tool does not currently have the functionality to adjust sampling weights to account for pooled samples. There are a few ways you can complete your analysis despite this limitation. You can analyze one sample per table cell. For example, you can analyze one month+year of BMS data at a time. You can accomplish this by using the row, column, and filter inputs strategically. Or you can analyze one year of ASEC data at a time (the ASEC only happens once per year, so if you are analyzing a single year of ASEC data, you do not need to worry about pooled samples). Another alternative is to create a microdata extract from IPUMS CPS and analyze the data using a statistical software package. Please let me know if you would like any guidance on that.
Below is how I would approach your question using the online data analysis tool for the year 2022.
Your current filters restrict the table to people whose hourly wage is between $1 and $15, who are the head of their household (PERNUM=1), and who live in Maryland.
The following input (using ASEC samples) will result in a table of the number of people whose wage is between $0.10 and $15.00 by state in the year 2022:
In 2022, I see that about 929,000 workers in Maryland who are in the universe of HOURWAGE earned $15.00 or less per hour. That amounts to about 29% of Maryland workers that year. This does not seem obviously incorrect to me.
Note that I’ve selected EARNWT - Earnings Weight in the Weight dropdown menu. Since the variable HOURWAGE is part of the Earner Study, also called the Outgoing Rotation Group, you must use EARNWT to weight this analysis. You can read more about weights here.
Also note that HOURWAGE is only available through March of 2023. HOURWAGE2, a rounded version of HOURWAGE, is available to use instead. When you analyze 2023 data, you’ll need to choose between using HOURWAGE (and therefore not using samples from April or later), versus using HOURWAGE2. You may want to use HOURWAGE2 for all of your analysis, since it is available in all the samples, and will improve the comparability of your estimates over time.
