Two Rent Estimates: Work Through the Comparison

You’re comparing two areas before looking at apartments. One summary shows $1,400 a month; the other shows $1,520. What would you need before writing, “The second area has higher rent”?

Work through the invented comparison below. First examine the two ranges, then calculate the difference, change one input and check your conclusion. None of these amounts describes an actual place, survey release or available apartment.

Rent estimate comparison: the fictional packet

Imagine two classroom surveys estimating the same quantity: mean monthly rent excluding utilities for occupied rental homes. Both use the same reference period, currency and dollar basis, with otherwise comparable population definitions. Call the fictional areas A and B.

For this exercise, assume the estimates have adequately normal, bell-shaped sampling distributions and independent estimation errors. Those assumptions are supplied by the exercise. Different area names alone wouldn’t establish them.

The four amounts are exact classroom inputs. Each margin is a finite, strictly positive number. Missing values, zero margins and special survey codes are outside this worksheet. We aren’t deriving a mean or its margin from individual rent records.

Draw the ranges, but leave the verdict blank

Subtract and add $80 around each estimate:

• A: $1,400 − $80 to $1,400 + $80, giving $1,320–$1,480.
• B: $1,520 − $80 to $1,520 + $80, giving $1,440–$1,600.

The ranges overlap between $1,440 and $1,480. Write that observation down without yet deciding whether the difference passes a statistical test.

The Census ACS handbook, chapter 7, warns against using overlapping individual confidence intervals as a significance test.

Calculate the difference separately

B minus A is $1,520 − $1,400 = $120. Keep that difference separate from either area’s individual margin.

For the stipulated 90% margins, divide each margin by 1.645 to obtain its standard error. See the Census statistical-testing instructions; this worksheet doesn’t cover older ACS conversion conventions or special missing-value codes.

Square the two standard errors, add them, then take the square root. Divide the estimate difference by that result. For a two-sided test, its absolute value must exceed 1.645 at 90% confidence or 1.96 at 95%. Equality doesn’t pass.

1. Each standard error: $80 ÷ 1.645 ≈ $48.632.
2. Combined standard error: square root of [($80 ÷ 1.645)² + ($80 ÷ 1.645)²] ≈ $68.776.
3. Test value: $120 ÷ $68.776 ≈ 1.745.

The displayed decimals are shortened; the underlying calculation retains the exact input values. At the two predeclared levels, 1.745 is above 1.645 and below 1.96. The same fictional packet therefore passes the 90% test and doesn’t pass the 95% test.

Here is an exact check of the 90% result using these inputs: $120² = 14,400, while $80² + $80² = 12,800. The first is larger. The model uses exact comparisons, so a rounded display cannot silently turn equality into a passing result.

Return to your overlapping ranges. They were calculated correctly, yet the baseline comparison passes the 90% test. That’s the specific counterexample this exercise was designed to reveal.

A statistically significant result concerns evidence of a population difference under the test’s assumptions. A result that isn’t significant doesn’t prove equality. These calculations address sampling uncertainty, not every possible survey error.

Rent estimate comparison: change only B’s estimate

In a separate hypothetical packet, replace B’s $1,520 with $1,500. Leave A and both $80 margins unchanged. This is a new classroom scenario, not a later observation or a measured rent decline.

The difference is now $100. The combined standard error stays about $68.776, making the test value about 1.454. This packet doesn’t pass either declared test. Its exact 90% check is 10,000 versus 12,800.

Keep both levels visible. Don’t select whichever produces your preferred conclusion. For the second packet, record “difference unresolved by these tests,” rather than writing that the areas have identical rents.

Know when this worksheet can’t answer

Now remove B’s margin from the packet. You can still subtract the rent estimates, but you can’t complete this worksheet’s uncertainty calculation. “Unavailable” is a different outcome from a completed test that doesn’t pass.

Check the packet manually before doing the arithmetic. Stop this exercise if:

• One figure is a mean and the other a median, or they measure different costs.
• The reference periods, currencies or dollar bases differ.
• The population scopes don’t match, such as all occupied rental homes versus newly advertised vacant homes.
• Observations overlap or repeat, estimation errors are dependent, or independence isn’t established.
• The normal-approximation assumption is absent or unjustified.
• The margins have different confidence bases, are missing, or fail the positive-number requirement.

These are the boundaries of our invented exercise, not a complete ACS rulebook. The arithmetic doesn’t inspect metadata or validate the assumptions. Changing one of these conditions can make the worksheet unsuitable even when all four amounts remain unchanged.

For actual ACS comparisons, rounded published estimates and margins can produce a different result from official tests using unrounded values.

Carry the result into your research, not your rent budget

Save the input packet beside your calculation. Write which fictional scenario you used, both results, and which assumptions you checked. That makes your conclusion inspectable instead of leaving a bare “higher” label.

For an apartment search, build a separate list of actual asking rents, utilities, deposits, travel costs and your own payment room. This exercise supplies none of those figures. It doesn’t say how much you can afford, predict an individual apartment’s price or explain why any real area costs more.

Original hypothetical research-reading exercise. Official methodological references checked October 21, 2026. No current rental data or Census tool output is represented here.