1800 Census of Albany Visualization

1800 Census of Albany Visualization

by Ryan Scarpelli

I. Data Description

For the intents and purposes of the final assignment in this course, I have chosen the 1800 Census of Albany, New York as my dataset. The reason I chose this specific set of data is three-fold. First, I was particularly interested by the age of the census. The 1800 Census was only the second of its kind taken in Albany, the first was taken 10 years prior in 1790 (Grondahl). Second, because this data is so historical it contains records of slave-ownership and the race/gender of the head of household. This will make it rather rewarding to work with because there is not much currently known about slavery in Albany, NY during this period. Lastly, I am a current resident in the City of Albany so the data is of particular personal interest.

The raw data for the 1800 Census can be found on the course resources page and is downloadable as a Comma-Separated Values (.CSV) file. The file contains a spreadsheet which contains 946 total records. Each record is a household that was surveyed during the census. For each record we are given 16 different columns of quantitative and qualitative data used to describe the homeowner, the home, and the occupants. The 16 columns are as follows:

  1. Full Name – A string value containing the given name and surname of the homeowner.
  2. Head of Household Race – A string value, either “White” or “Black”, which describes the race of the homeowner.
  3. Head of Household Gender – A string value, either “Male” or “Female”, which describes the gender of the homeowner.
  4. Ward Number – A number value (1, 2, or 3) which corresponds to the area within Albany the homeowner lives.
  5. Free White Men < 10 Years – A numerical headcount for Free White Men within the household. (Applies to all Free White Men columns)
  6. Free White Men 10-16 Years
  7. Free White Men 16-26 Years
  8. Free White Men 26-45 Years
  9. Free White Men > 45 Years
  10. Free White Women < 10 Years – A numerical headcount for Free White Women within the household. (Applies to all Free White Women columns)
  11. Free White Women 10-16 Years
  12. Free White Women 16-26 Years
  13. Free White Women 26-45 Years
  14. Free White Women > 45 Years
  15. Other Free Persons – A numerical headcount for persons described as “other…excepting Indians not taxed” within the household.
  16. Slaves – A numerical headcount for enslaved persons within the household.

For the sake of cleaning up our data for the uses of this assignment, I made a few alteration to the spreadsheet. First, I created a new column simply called “Free White Men” which aggregated the total Free White Men values for all ages into one column. This essentially condenses the records on Free White Men for each individual household. Next, I followed the same exact steps to create an aggregate column called Free White Women. Third, once both of these columns were created I deleted the columns I had just aggregated; I no longer need them as I now have their data in the respective totals column. The results are that we now have 8 columns, half of how me we had at the start.

CustomWardMapOverlay
Custom Ward Map Overlay (Click to Enlarge)

After alterations were made, I performed my own independent research on the Wards column. Our dataset contains a Wards column, with the only possible values being the numbers 1, 2, and 3. I found that these numbers correspond to political subdivision of Albany which were implemented in 1790 (Bielinski). The issue with these values is that ward lines for the City of Albany have drastically changed over time. Today, Albany is currently divided into 16 of these wards and none match the wards referenced in the census. Thankfully, I was able to find a description of the ward lines including very dated maps on the New York State Museum website (Bielinski). The next issue I found was that the historical map was very dated and illegible, I could not meaningfully draw any conclusions about the wards. Stuck with a currently-useless map, I turned to Adobe Photoshop CC to see if I could manipulate the image for my needs. First, I captured a screenshot of a map of Albany, NY from Google Maps and imported it. Next, I imported the image of the historical map on top of the Google Map. From there, I reduced the opacity of the historical so that I could see the current map underneath. Using the transform tool on the historical map, I was able to rotate it to align the two maps perfectly. The new hybrid map made it much easier to identify physical features, street names, and ward lines. The final step was to find which geographic coordinates and zip codes correspond to the wards I had just identified. After a quick internet search, I added these values to 3 new columns: Ward Latitude, Ward Longitude, and Ward Zip.

My intention for this dataset is to manipulate the data and make comparisons between values. In doing so, I hope to be able to identify discernible trends between the Homeowner Race, Homeowner Gender, and the types of occupants within the household.

II. Data Visualizations

 

Visualization 1 – House Demographics

For the first visualization, I started by introducing my data and findings at the most general level. Being that this dataset is from the 1800 Census, I began with comparisons about the households themselves before diving into the occupants/population. There were 946 total houses surveyed during this census. I was highly interested in seeing how that number breaks down by race and gender independently before comparing them simultaneously. Therefore, I created three independent graphics: a breakdown of houses by head race, a breakdown of houses by head gender, and a simultaneous breakdown of houses by both head race and gender.

  • The first pie chart breaks down the totality of houses by the race of their head, either White (light green) or Black (dark green). Out of the total 946 houses, 914 were run by a White head and only a miniscule 32 were run by a Black head of household. This breakdown is staggering as houses with a White head of household account for over 96.6% of all homes recorded.
  • The second pie chart is very similar to the first; however, the totality of houses is now broken down from 946 by gender (instead of race) of the head. As you can see, there were 853 Male-head houses (blue) and 93 Female-head houses (purple). This breakdown shows a marginally smaller differential than the race comparison, however Male-led houses still represented 90.1% of the population.
  • The final graphic for this visualization is a packed bubbles chart which divides the households in four head-types: White Male, White Female, Black Male, and Black Female. Basically, this chart combines the findings of the two previous charts to see a more accurate depiction of how many homes were run by each of the types of head. The cumulative breakdown of the 946 homes shows that of the four head-types, the predominant leader is White Males with a count of 833. Inversely, Black Females were the smallest group with a total count of only 12 houses.

 

Visualization 2 – Race vs. Gender (Totals)

Now that we have a solid understanding of the types of houses we are working with at general level, we are going to look into meaningful trends about the occupants of these houses. As stated above, our dataset broke down occupants into four main categories: Free White Males (blue), Free White Females (purple), Other Free Persons (orange), and Slaves (green). To better understand what effect the head-type had on these person counts, I divided homes once again by head race and head gender to compare. This visualization contains two independent graphs.

  • The first graph compares the Race of the Head of Household to the Breakdown of Occupants. Basically, I am comparing the race of the head, white or black, to the number of occupants in each category. We found out in the previous visualization that the number of White Households (914) heavily outweighs the number of Black Households (32). This fact has carried through and is evident in this graph as well.The results of this comparison are that White Households lead Black Households in every category except Other Free Persons. In the case of Other Free Persons, White Households contain 54 people while Black Households contain 95 people. Additionally, the 95 Other Free Persons in Black Households account for more than double the other counts combined. Other interesting aspects include the fact that Black Households in total only had 2 Free White Men and 2 Free White Women. The historical context of this data correlates with the findings.
  • The second bar graph in the visualization compares the Gender of the Head of Household to the Breakdown of Occupants. As you can see, the comparison between White-Black and Male-Female Households is very similar. The Male Households beat out the Female Households across the board in each category.

Visualization 3 – Average Occupancy by Race/Gender

The third visualization displays the average number of Free White Men, Free White Women, Other Free Persons, and Slaves when filtered by Race, Gender, and both. The intention is to compare the averages to the head-type to identify any discernable trends. The first graph will show the average occupancies when divided by Race. The second will show the same comparison but by Gender. The third graph will compound the findings of the previous graphs into one seamless table.

  • The first graph displays the average counts for each person-type by the race of the head of household. The key finding in this bar chart is that Black Households contain on average 2.969 Other Free Persons while White Household only contain 0.059. This means that on average there are nearly 3 (2.91) additional Other Free Persons in Black Households which represents an increase of a whopping 4932.20%. A shocking value to look at is that Black Households led in Average Slaves with 0.781 to White’s 0.562. This may be counterintuitive to some to see that Black Households had more slaves on average.
  • The second graph displays the average counts for each person-type by the gender of the head of household. The results from this graph do not indicate any logical upsets; the findings are exactly what you would expect from our previous data. Male Households beat Female Households in average count for each person-type category.
  • The third graph shows the data from the previous graphs combined. Therefore, it displays the average counts for each person-type by race and gender. This is by far my favorite chart from the entire assignment as I believe it is the most insightful. Our whole population is broken this time by Race AND Gender to display the average counts for each person-type.

There are a number of important and interesting results from the final graph. First, Black-Female Households contained the highest average number of slaves (1.083) while White-Male contained the lowest (0.551). Second, Black-Male Households contained the most average Other Free Persons (3.450) while White-Male households had the lowest (0.059)

Visualization 4 – Slave Distribution

The fourth visualization consists of three packed bubbles graphs which portray homeowners with the most amount of slaves.

The first chart shows the homeowners with the highest number of slaves. I chose a packed bubbles graph because it is easy to view large amounts of data and also because the attributes of size and color allow me further make points. Chart 4A was constructed by setting the Name dimension as a column and the sum of the Slaves measure as a row. I was then able to filter the results to only show homeowners with at least once slave. Next, I used the sum of the Slaves measure to indicate both color and size. This allows us to nicely differentiate and find the houses with the highest slave population.

The second and third charts (4B – Highest Slaves by Gender and 4C – Highest Slaves by Race) are very similar to one another. Both of these charts are packed bubble graphs which are used to represent the same data as 4A. The difference is that for 4B, Head of Household Gender is used as the color indicator instead of Slaves. This allows us to easily see which homes are male-run or female-run, and where the highest concentration of slaves lies (hint: Males). On the other hand, Chart 4C does the same exact thing however this time we are differentiating by race instead of gender.

 

Visualization 5 – Wards

The fifth and final visualization is the results of my pain-staking work with Albany’s wards. Using the reference from the NYSM website, I was able to recreate the ward lines from 1800 on a current map of Albany. From there, I found their geographic coordinates online. Lastly, I researched which current Albany zip codes most closely match up to the historical ward lines. Though the match is not perfect, it is close enough to be a decently accurate representation. The visualization consists of two filled-map graphs which display the number of homes and people in each ward, respectively.

The first graph, 5A, uses a filled-map chart to show the number of homes in each of the three wards. It was created by importing the custom values for zip codes I found. Once the zip codes were part of our data, I assigned the geographic role “Zip Code” to the measure Ward Zip before assigning the measure as a column. This nicely produced a rough version of the map you are looking at. I assigned the sum of the Number of Records as the color differentiator; higher populations are darker blue, lower populations are lighter blue. Finally, I edited the label to include the Ward Number and the Number of People in each ward.

The second graph, 5B, is almost identical to 5A. It only differs in that the colors represent the number of individual people in each ward instead of households. Please see above for further detail.

 

 

 

 

III. Process Documentation

The first visualization is intended to introduce the audience to the number of households by Race and Gender.

  1. The comparison I chose was White Households vs. Black Households in order to show the large discrepancy in number between the two. I created this visualization by assigning the dimension Head of Household Race to the column and Number of Records to the rows. Due to the fact that we are comparing sections of a population to the whole, it was an easy choice to select a bar graph as my graph-type. When working with visuals representing race, I find it the least-awkward to use shades of the same color rather than randomly assigning colors. This helps avoid assigning specific colors to each race which some may find offensive.
  2. The comparison being made is the number of Male-Head Households vs. Female-Head Households. This graph was constructed by placing Head of Household Gender in the column and Number of Records in the row. Along the same vein as Chart A, I chose the pie chart style to represent chunks of a whole. I selected traditional colors which correspond to gender: Blue for Males, Purple for Females. My choices for this graph clearly highlight the drastic difference in number of Male Household over Female Households.
  3. Chart C takes a unique-but-similar approach to Chart’s A & B. It compares the number of households when simultaneously divided by race and gender into four categories: White Male, White Female, Black Male, and Black Female. To do this, I placed Head of Household Race and Head of Household Gender into the columns, and Number of Records into rows. I used a packed bubbles chart so that I could manipulate the appearance of each category by attributes of my choosing. I chose to have the Number of Records indicate Size, Color indicate Head of Household Gender, and I formatted the label to appear cleaner. I believe it makes it easier to visualize the data and its underlying meaning with these attributes.

The visualization itself consists of these three graphs organized into one dashboard. This provides a detailed view of the houses by their criteria.

The second visualization is intended to show the audience a deeper look into the census. In this visualization, we will be breaking down the total number of each person-type by the race and gender of the Head of Household. The reason I chose to compares these values is because it allows us to identify if the type of head affects the occupants of the household.

  1. The first graph for the second visualization compares the Head of Household Race to the Number of Occupants by Type. The intention in making this comparison is to see whether White/Black Households are more/less likely to contain a certain type of person. I chose a side-by-side bar graph to display these totals so that their values can be easily compares. I assigned Head of Household Race to the column and the sum of measures Free White Men, Free White Women, Other Free Persons, and Slaves to the row. Again, I chose Blue for Men, Purple for Women, and this time I assigned Orange to Other Free Persons and Green to Slaves.
  2. The second graph displays the effect of the Head of Household Gender of the total occupants in each person-type. To construct this graph, I dragged the dimension Head of Household Gender to columns and the sum of measures Free White Men, Free White Women, Other Free Persons, and Slaves to the rows. The colors match those in Graph A.

The visualization itself was designed by placing the graphs on top of each other within a custom dashboard. The key/legend for both graphs appears in between them.

The third and final visualization focuses on the average counts of each person-type by race and gender. The reason I chose to use averages is to get a clearer understanding of the data when dealing with big differences in sample size (Male-Female, White-Black).

  1. The first graph of the third visualization compares Household Race and Person-Type Averages. The process to create this graph included dragging the dimension Head of Household Race to the columns and the average of measures Free White Men, Free White Women, Other Free Persons, and Slaves to the rows.
  2. The second graph represents the comparison between Household Gender and Person-Type Averages. I took the following steps to make this graph: assign Head of Household Gender to columns, assign Free White Men, Free White Women, Other Free Persons, and Slaves to rows.
  3. The third graph combines the data from the previous two; it accurately depicts the relationship between Household Race, Household Gender, and Person-Type Averages. To combine the two previous graphs I assigned Head of Household Race and Head of Household Gender to columns. Next, I assigned the average of measures Free White Men, Free White Women, Other Free Persons, and Slaves to the rows.

The visualization was constructed by incorporating the three graphs into a custom dashboard. The first two graphs appear side-by-side with the combined graph appearing below. The key appears in the middle of the graph.

IV. Argumentation

Argument A

When dealing with data from the past, especially from a census as far back as 1800, it is absolutely necessary to understand the historical context behind the information. The societal norms and expectations of this time over two centuries ago are a far cry from those familiar in 2016. Namely, the institution of slavery was still very much alive and profitable during this time. This census was only the second to ever be taken for the City of Albany; the first census occurred in 1790. The 1790 census revealed a staggering number of slaves within Albany – 3,722 to be exact out of 3,498 residents. This number ranked Albany first in the entire state for slave population in 1790 (Grondahl).

Therefore, there is no surprise that the very structure of the census data reflects slave ownership. During this time in Albany’s history, slaves were unfortunately extremely common. They were owned by residents ranging from minimal income-earners to prominent, posh families such as the Schuyler family of the Schuyler Mansion (Grondahl). Thankfully, our census only has records of 539 enslaved persons within Albany. While this number is a positive trend of decreasing slave numbers, it is also drastic enough to question the cause. Through online research I was able to find a possible cause of this drastic decrease, from our own Times Union (Grondahl):

The annuity program was put in place to compensate slave owners after the gradual manumission act was passed. The payments were for infants born to slaves after 1799 and they were placed on a permanent state poor list. Masters of each slave child received between $12 to $18 a year until the slave reached the age of 25. On average, the state paid each owner $400 to $500 per slave in annuities.

“The fear on the part of the government was that if you freed all the slaves immediately, they’d become a huge burden on the state,” Barbagallo said. “That just wasn’t the case. Those were misconceptions based on racism.”

Essentially, once manumission was passed within Albany it started a trend of large-scale offloading of slaves. The government feared a huge influx of recently-freed albeit penniless slaves so they began offering owners compensation for kin of slaves born after 1799 (Grondahl). This is one of the many factors that led to the dispersion of slaves from Albany. While this fact, which largely signaled the ‘beginning-of-the-end’ of slavery, heavily contributed to decreasing numbers it wasn’t the sole reason.

The other key determinants which led to a decreasing slave population had little to do with law and everything to do with location. Albany has historically been a travel and trade hotspot for the upper New York region, especially during this time period. When the 1800 census was taken, Albany had just completed construction on a turnpike which vastly opened up channels of travel to the capital city (Wiki). Moreover, by the year 1815 Albany had already become the turnpike center of New York State.  The high accessibility of the city combined with high slave population, increasing slave laws, and emerging contacts with the outside world may all explain the decrease.

Mini-Argument B

With respect to key findings in my analysis, I would like to focus on average slave occupancy by homeowner type. I have provided a comprehensive breakdown of how homeowner race/gender effects the types of occupants in their household. The results of my analysis were largely expected and/or predictable, however there were anomalies and findings regarding which I was surprised.

In Chart 3C “Average Occupants by Demographic”, we can clearly see that houses with Black Female homeowners contain the largest average number of slaves. For reference, the relevant data is below:

  • White Male – 0.551 Average Slaves
  • White Female – 0.679 Average Slaves
  • Black Male – 0.600 Average Slaves
  • Black Female – 1.083 Average Slaves

The chart states that by average slave ownership, from most to least is: Black Female, White Female, Black Male, White Male. This result is counterintuitive as most people would expect White Males to have the highest number, not the lowest. There are a number of possible causes for this ranking order.

For example, all counts for White Male Households may be skewed as they make up the overwhelming majority of all households. As a result, all averages are diluted when compared to Homeowners with miniscule numbers such as Black Females. I personally believe this is the most likely cause, however alternative explanations exist. Out of all households, only 94/946 were owned by Females and just 32/946 had a Black head of household. When combined, the number of Black AND Female homeowner is microscopic.

V. Further Research Questions

The primary question I would like to pose looking forward relates to the outward dispersion of both enslaved and recently-freed persons from Albany. In the decade from 1790 when the first census was taken until 1800, over 3,183 persons previously listed as slaves had left the city. The most basic question I have regarding this finding is “Where did they go?” There are a number of approaches I could take to find my answer.

First, I could perform traditional topic-area research. There are undoubtedly countless articles and journals about the migration of slaves in New York during this period. Particularly, I believe that the journal “Slavery in Albany, New York, 1624-1827” by Oscar Williams may cover this exact practice in detail (Williams). The reason I did not for this assignment is due to the journal being pay-to-access.

The second research method is more advanced. To truly find out where these people have gone after leaving Albany, I need to compare more data. For this assignment I examined a single spreadsheet containing data from a single year in a single county. In order to accurately track the missing persons, I would need to build a custom MySQL database. First, I would need to important all records from the 1790 Albany census as a baseline. I would only keep records on the persons listed as slaves, as they are only of interest. Second, I would import records for all Albany censuses until around 1915. This is done to track the slaves that never actually left the city. Finally, I would have to import records from all other counties within New York State beginning with 1790. Ultimately, using advanced database queries and perhaps some custom programming and scripting I would hopefully be able to track these persons and their location across time.

 

VI. References

Bielinski, Stefan. “Albany Wards.” Albany Wards. Accessed May 12, 2016. http://exhibitions.nysm.nysed.gov//albany/wards.html.

Grondahl, Paul. “Ultimate Payoff for Slaves’ Freedom.” Times Union. Accessed May 12, 2016. http://www.timesunion.com/local/article/Ultimate-payoff-for-slaves-freedom-3684612.php.

Grondahl, Paul. “Albany’s Long, Neglected History of Slavery.” Times Union. Accessed May 12, 2016. http://www.timesunion.com/tuplus-local/article/Albany-s-long-neglected-history-of-slavery-6808527.php.

Oscar, William. “Slavery in Albany, New York, 1624-1827.” Slavery in Albany, New York by Williams, Oscar. Accessed May 12, 2016. https://www.questia.com/library/journal/1G1-236731633/slavery-in-albany-new-york-1624-1827.

“History of Albany, New York.” Wikipedia. Accessed May 12, 2016. https://en.wikipedia.org/wiki/History_of_Albany,_New_York.